├── .github ├── FUNDING.yml ├── ISSUE_TEMPLATE │ └── bug-report.md └── pictures │ └── alipay.JPG ├── .gitignore ├── CITATION.cff ├── LICENSE ├── README.md ├── docs ├── Author.md ├── Changelog.md ├── Install.md ├── Makefile ├── Quickstart.md ├── Recommend.md ├── State.md ├── conf.py ├── index.rst ├── logo.png ├── make.bat ├── pikachu.jpg ├── requirements.txt ├── screenshot_characterize.gif ├── screenshot_fastneuralstyletransfer.gif ├── screenshot_photomosaic.png └── screeshot_noteprocessor.png ├── pydrawing ├── __init__.py ├── asserts │ ├── animecharacter.jpg │ ├── badapple.mp4 │ ├── badapple_color.mp4 │ ├── bridge.png │ ├── dog.jpg │ ├── face.jpg │ ├── monalisa.jpg │ ├── note.jpg │ └── zurich.jpg ├── modules │ ├── __init__.py │ ├── beautifiers │ │ ├── __init__.py │ │ ├── base │ │ │ ├── __init__.py │ │ │ └── base.py │ │ ├── beziercurve │ │ │ ├── __init__.py │ │ │ ├── beziercurve.py │ │ │ └── potrace.exe │ │ ├── cartoongan │ │ │ ├── __init__.py │ │ │ └── cartoongan.py │ │ ├── cartoonise │ │ │ ├── __init__.py │ │ │ └── cartoonise.py │ │ ├── cartoonizeface │ │ │ ├── __init__.py │ │ │ ├── cartoonizeface.py │ │ │ ├── facedetector │ │ │ │ ├── __init__.py │ │ │ │ └── facedetector.py │ │ │ └── facesegmentor │ │ │ │ ├── __init__.py │ │ │ │ └── facesegmentor.py │ │ ├── characterize │ │ │ ├── __init__.py │ │ │ └── characterize.py │ │ ├── douyineffect │ │ │ ├── __init__.py │ │ │ └── douyineffect.py │ │ ├── fastneuralstyletransfer │ │ │ ├── __init__.py │ │ │ └── fastneuralstyletransfer.py │ │ ├── geneticfitting │ │ │ ├── __init__.py │ │ │ ├── geneticfittingcircle.py │ │ │ └── geneticfittingpolygon.py │ │ ├── glitch │ │ │ ├── __init__.py │ │ │ └── glitch.py │ │ ├── nostalgicstyle │ │ │ ├── __init__.py │ │ │ └── nostalgicstyle.py │ │ ├── noteprocessor │ │ │ ├── __init__.py │ │ │ └── noteprocessor.py │ │ ├── oilpainting │ │ │ ├── __init__.py │ │ │ └── oilpainting.py │ │ ├── pencildrawing │ │ │ ├── __init__.py │ │ │ ├── pencildrawing.py │ │ │ └── textures │ │ │ │ └── default.jpg │ │ ├── photocorrection │ │ │ ├── __init__.py │ │ │ └── photocorrection.py │ │ └── photomosaic │ │ │ ├── __init__.py │ │ │ └── photomosaic.py │ └── utils │ │ ├── __init__.py │ │ ├── io.py │ │ └── logger.py └── pydrawing.py ├── requirements.txt └── setup.py /.github/FUNDING.yml: -------------------------------------------------------------------------------- 1 | patreon: CharlesPikachu 2 | ko_fi: charlespikachu 3 | custom: https://github.com/CharlesPikachu/Games/tree/master/.github/pictures/alipay.JPG 4 | -------------------------------------------------------------------------------- /.github/ISSUE_TEMPLATE/bug-report.md: -------------------------------------------------------------------------------- 1 | --- 2 | name: Bug report 3 | about: What bug did you encounter 4 | title: "[BUG]" 5 | labels: '' 6 | assignees: '' 7 | --- 8 | 9 | **Environment (使用环境)** 10 | 11 | - Installation method (安装方式): 12 | - The version of pydrawing (版本号): 13 | - Operating system (操作系统): 14 | - Python version (Python版本): 15 | 16 | **Question description (问题描述)** 17 | 18 | **Screenshot (报错截图)** 19 | 20 | **Advice (修复建议)** 21 | -------------------------------------------------------------------------------- /.github/pictures/alipay.JPG: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/CharlesPikachu/pydrawing/e980ad9bf4cece42ff40ed2bc7bed7a155eabb8e/.github/pictures/alipay.JPG -------------------------------------------------------------------------------- /.gitignore: -------------------------------------------------------------------------------- 1 | # Byte-compiled / optimized / DLL files 2 | __pycache__/ 3 | *.py[cod] 4 | *$py.class 5 | 6 | # C extensions 7 | *.so 8 | 9 | # Distribution / packaging 10 | .Python 11 | build/ 12 | develop-eggs/ 13 | dist/ 14 | downloads/ 15 | eggs/ 16 | .eggs/ 17 | lib/ 18 | lib64/ 19 | parts/ 20 | sdist/ 21 | var/ 22 | wheels/ 23 | *.egg-info/ 24 | .installed.cfg 25 | *.egg 26 | MANIFEST 27 | 28 | # PyInstaller 29 | # Usually these files are written by a python script from a template 30 | # before PyInstaller builds the exe, so as to inject date/other infos into it. 31 | *.manifest 32 | *.spec 33 | 34 | # Installer logs 35 | pip-log.txt 36 | pip-delete-this-directory.txt 37 | 38 | # Unit test / coverage reports 39 | htmlcov/ 40 | .tox/ 41 | .coverage 42 | .coverage.* 43 | .cache 44 | nosetests.xml 45 | coverage.xml 46 | *.cover 47 | .hypothesis/ 48 | .pytest_cache/ 49 | 50 | # Translations 51 | *.mo 52 | *.pot 53 | 54 | # Django stuff: 55 | *.log 56 | local_settings.py 57 | db.sqlite3 58 | 59 | # Flask stuff: 60 | instance/ 61 | .webassets-cache 62 | 63 | 64 | # Scrapy stuff: 65 | .scrapy 66 | 67 | # Sphinx documentation 68 | docs/_build/ 69 | 70 | # PyBuilder 71 | target/ 72 | 73 | # Jupyter Notebook 74 | .ipynb_checkpoints 75 | 76 | # pyenv 77 | .python-version 78 | 79 | # celery beat schedule file 80 | celerybeat-schedule 81 | 82 | # SageMath parsed files 83 | *.sage.py 84 | 85 | # Environments 86 | .env 87 | .venv 88 | env/ 89 | venv/ 90 | ENV/ 91 | env.bak/ 92 | venv.bak/ 93 | 94 | # Spyder project settings 95 | .spyderproject 96 | .spyproject 97 | 98 | # Rope project settings 99 | .ropeproject 100 | 101 | # mkdocs documentation 102 | /site 103 | 104 | # mypy 105 | .mypy_cache/ 106 | 107 | -------------------------------------------------------------------------------- /CITATION.cff: -------------------------------------------------------------------------------- 1 | cff-version: 1.2.0 2 | message: "If you use this software, please cite it as below." 3 | authors: 4 | - name: "Zhenchao Jin" 5 | title: "Pydrawing: Beautify your image or video" 6 | date-released: 2022-01-18 7 | url: "https://github.com/CharlesPikachu/pydrawing" 8 | license: Apache-2.0 9 | -------------------------------------------------------------------------------- /LICENSE: -------------------------------------------------------------------------------- 1 | Apache License 2 | Version 2.0, January 2004 3 | http://www.apache.org/licenses/ 4 | 5 | TERMS AND CONDITIONS FOR USE, REPRODUCTION, AND DISTRIBUTION 6 | 7 | 1. Definitions. 8 | 9 | "License" shall mean the terms and conditions for use, reproduction, 10 | and distribution as defined by Sections 1 through 9 of this document. 11 | 12 | "Licensor" shall mean the copyright owner or entity authorized by 13 | the copyright owner that is granting the License. 14 | 15 | "Legal Entity" shall mean the union of the acting entity and all 16 | other entities that control, are controlled by, or are under common 17 | control with that entity. For the purposes of this definition, 18 | "control" means (i) the power, direct or indirect, to cause the 19 | direction or management of such entity, whether by contract or 20 | otherwise, or (ii) ownership of fifty percent (50%) or more of the 21 | outstanding shares, or (iii) beneficial ownership of such entity. 22 | 23 | "You" (or "Your") shall mean an individual or Legal Entity 24 | exercising permissions granted by this License. 25 | 26 | "Source" form shall mean the preferred form for making modifications, 27 | including but not limited to software source code, documentation 28 | source, and configuration files. 29 | 30 | "Object" form shall mean any form resulting from mechanical 31 | transformation or translation of a Source form, including but 32 | not limited to compiled object code, generated documentation, 33 | and conversions to other media types. 34 | 35 | "Work" shall mean the work of authorship, whether in Source or 36 | Object form, made available under the License, as indicated by a 37 | copyright notice that is included in or attached to the work 38 | (an example is provided in the Appendix below). 39 | 40 | "Derivative Works" shall mean any work, whether in Source or Object 41 | form, that is based on (or derived from) the Work and for which the 42 | editorial revisions, annotations, elaborations, or other modifications 43 | represent, as a whole, an original work of authorship. For the purposes 44 | of this License, Derivative Works shall not include works that remain 45 | separable from, or merely link (or bind by name) to the interfaces of, 46 | the Work and Derivative Works thereof. 47 | 48 | "Contribution" shall mean any work of authorship, including 49 | the original version of the Work and any modifications or additions 50 | to that Work or Derivative Works thereof, that is intentionally 51 | submitted to Licensor for inclusion in the Work by the copyright owner 52 | or by an individual or Legal Entity authorized to submit on behalf of 53 | the copyright owner. For the purposes of this definition, "submitted" 54 | means any form of electronic, verbal, or written communication sent 55 | to the Licensor or its representatives, including but not limited to 56 | communication on electronic mailing lists, source code control systems, 57 | and issue tracking systems that are managed by, or on behalf of, the 58 | Licensor for the purpose of discussing and improving the Work, but 59 | excluding communication that is conspicuously marked or otherwise 60 | designated in writing by the copyright owner as "Not a Contribution." 61 | 62 | "Contributor" shall mean Licensor and any individual or Legal Entity 63 | on behalf of whom a Contribution has been received by Licensor and 64 | subsequently incorporated within the Work. 65 | 66 | 2. Grant of Copyright License. Subject to the terms and conditions of 67 | this License, each Contributor hereby grants to You a perpetual, 68 | worldwide, non-exclusive, no-charge, royalty-free, irrevocable 69 | copyright license to reproduce, prepare Derivative Works of, 70 | publicly display, publicly perform, sublicense, and distribute the 71 | Work and such Derivative Works in Source or Object form. 72 | 73 | 3. Grant of Patent License. Subject to the terms and conditions of 74 | this License, each Contributor hereby grants to You a perpetual, 75 | worldwide, non-exclusive, no-charge, royalty-free, irrevocable 76 | (except as stated in this section) patent license to make, have made, 77 | use, offer to sell, sell, import, and otherwise transfer the Work, 78 | where such license applies only to those patent claims licensable 79 | by such Contributor that are necessarily infringed by their 80 | Contribution(s) alone or by combination of their Contribution(s) 81 | with the Work to which such Contribution(s) was submitted. If You 82 | institute patent litigation against any entity (including a 83 | cross-claim or counterclaim in a lawsuit) alleging that the Work 84 | or a Contribution incorporated within the Work constitutes direct 85 | or contributory patent infringement, then any patent licenses 86 | granted to You under this License for that Work shall terminate 87 | as of the date such litigation is filed. 88 | 89 | 4. Redistribution. You may reproduce and distribute copies of the 90 | Work or Derivative Works thereof in any medium, with or without 91 | modifications, and in Source or Object form, provided that You 92 | meet the following conditions: 93 | 94 | (a) You must give any other recipients of the Work or 95 | Derivative Works a copy of this License; and 96 | 97 | (b) You must cause any modified files to carry prominent notices 98 | stating that You changed the files; and 99 | 100 | (c) You must retain, in the Source form of any Derivative Works 101 | that You distribute, all copyright, patent, trademark, and 102 | attribution notices from the Source form of the Work, 103 | excluding those notices that do not pertain to any part of 104 | the Derivative Works; and 105 | 106 | (d) If the Work includes a "NOTICE" text file as part of its 107 | distribution, then any Derivative Works that You distribute must 108 | include a readable copy of the attribution notices contained 109 | within such NOTICE file, excluding those notices that do not 110 | pertain to any part of the Derivative Works, in at least one 111 | of the following places: within a NOTICE text file distributed 112 | as part of the Derivative Works; within the Source form or 113 | documentation, if provided along with the Derivative Works; or, 114 | within a display generated by the Derivative Works, if and 115 | wherever such third-party notices normally appear. The contents 116 | of the NOTICE file are for informational purposes only and 117 | do not modify the License. You may add Your own attribution 118 | notices within Derivative Works that You distribute, alongside 119 | or as an addendum to the NOTICE text from the Work, provided 120 | that such additional attribution notices cannot be construed 121 | as modifying the License. 122 | 123 | You may add Your own copyright statement to Your modifications and 124 | may provide additional or different license terms and conditions 125 | for use, reproduction, or distribution of Your modifications, or 126 | for any such Derivative Works as a whole, provided Your use, 127 | reproduction, and distribution of the Work otherwise complies with 128 | the conditions stated in this License. 129 | 130 | 5. Submission of Contributions. Unless You explicitly state otherwise, 131 | any Contribution intentionally submitted for inclusion in the Work 132 | by You to the Licensor shall be under the terms and conditions of 133 | this License, without any additional terms or conditions. 134 | Notwithstanding the above, nothing herein shall supersede or modify 135 | the terms of any separate license agreement you may have executed 136 | with Licensor regarding such Contributions. 137 | 138 | 6. Trademarks. This License does not grant permission to use the trade 139 | names, trademarks, service marks, or product names of the Licensor, 140 | except as required for reasonable and customary use in describing the 141 | origin of the Work and reproducing the content of the NOTICE file. 142 | 143 | 7. Disclaimer of Warranty. Unless required by applicable law or 144 | agreed to in writing, Licensor provides the Work (and each 145 | Contributor provides its Contributions) on an "AS IS" BASIS, 146 | WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or 147 | implied, including, without limitation, any warranties or conditions 148 | of TITLE, NON-INFRINGEMENT, MERCHANTABILITY, or FITNESS FOR A 149 | PARTICULAR PURPOSE. You are solely responsible for determining the 150 | appropriateness of using or redistributing the Work and assume any 151 | risks associated with Your exercise of permissions under this License. 152 | 153 | 8. Limitation of Liability. In no event and under no legal theory, 154 | whether in tort (including negligence), contract, or otherwise, 155 | unless required by applicable law (such as deliberate and grossly 156 | negligent acts) or agreed to in writing, shall any Contributor be 157 | liable to You for damages, including any direct, indirect, special, 158 | incidental, or consequential damages of any character arising as a 159 | result of this License or out of the use or inability to use the 160 | Work (including but not limited to damages for loss of goodwill, 161 | work stoppage, computer failure or malfunction, or any and all 162 | other commercial damages or losses), even if such Contributor 163 | has been advised of the possibility of such damages. 164 | 165 | 9. Accepting Warranty or Additional Liability. While redistributing 166 | the Work or Derivative Works thereof, You may choose to offer, 167 | and charge a fee for, acceptance of support, warranty, indemnity, 168 | or other liability obligations and/or rights consistent with this 169 | License. However, in accepting such obligations, You may act only 170 | on Your own behalf and on Your sole responsibility, not on behalf 171 | of any other Contributor, and only if You agree to indemnify, 172 | defend, and hold each Contributor harmless for any liability 173 | incurred by, or claims asserted against, such Contributor by reason 174 | of your accepting any such warranty or additional liability. 175 | 176 | END OF TERMS AND CONDITIONS 177 | 178 | APPENDIX: How to apply the Apache License to your work. 179 | 180 | To apply the Apache License to your work, attach the following 181 | boilerplate notice, with the fields enclosed by brackets "[]" 182 | replaced with your own identifying information. (Don't include 183 | the brackets!) The text should be enclosed in the appropriate 184 | comment syntax for the file format. We also recommend that a 185 | file or class name and description of purpose be included on the 186 | same "printed page" as the copyright notice for easier 187 | identification within third-party archives. 188 | 189 | Copyright [yyyy] [name of copyright owner] 190 | 191 | Licensed under the Apache License, Version 2.0 (the "License"); 192 | you may not use this file except in compliance with the License. 193 | You may obtain a copy of the License at 194 | 195 | http://www.apache.org/licenses/LICENSE-2.0 196 | 197 | Unless required by applicable law or agreed to in writing, software 198 | distributed under the License is distributed on an "AS IS" BASIS, 199 | WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. 200 | See the License for the specific language governing permissions and 201 | limitations under the License. 202 | -------------------------------------------------------------------------------- /README.md: -------------------------------------------------------------------------------- 1 |
2 | 3 |
4 |
5 | 6 | [![docs](https://img.shields.io/badge/docs-latest-blue)](https://pydrawing.readthedocs.io/) 7 | [![PyPI - Python Version](https://img.shields.io/pypi/pyversions/pydrawing)](https://pypi.org/project/pydrawing/) 8 | [![PyPI](https://img.shields.io/pypi/v/pydrawing)](https://pypi.org/project/pydrawing) 9 | [![license](https://img.shields.io/github/license/CharlesPikachu/pydrawing.svg)](https://github.com/CharlesPikachu/pydrawing/blob/master/LICENSE) 10 | [![PyPI - Downloads](https://pepy.tech/badge/pydrawing)](https://pypi.org/project/pydrawing/) 11 | [![PyPI - Downloads](https://img.shields.io/pypi/dm/pydrawing?style=flat-square)](https://pypi.org/project/pydrawing/) 12 | [![issue resolution](https://isitmaintained.com/badge/resolution/CharlesPikachu/pydrawing.svg)](https://github.com/CharlesPikachu/pydrawing/issues) 13 | [![open issues](https://isitmaintained.com/badge/open/CharlesPikachu/pydrawing.svg)](https://github.com/CharlesPikachu/pydrawing/issues) 14 | 15 | Documents: https://pydrawing.readthedocs.io/ 16 | 17 | 18 | # Pydrawing 19 | ``` 20 | Beautify your image or video. 21 | You can star this repository to keep track of the project if it's helpful for you, thank you for your support. 22 | ``` 23 | 24 | 25 | # Support List 26 | | Beautifier_EN | Introduction | Related Paper | Code | Beautifier_CN | 27 | | :----: | :----: | :----: | :----: | :----: | 28 | | glitch | [click](https://mp.weixin.qq.com/s/Yv0uPLsTGwVnj_PKqYCmAw) | N/A | [click](./pydrawing/modules/beautifiers/glitch) | 信号故障特效 | 29 | | douyineffect | [click](https://mp.weixin.qq.com/s/RRnrO2H84pvtUdDsAYD9Qg) | N/A | [click](./pydrawing/modules/beautifiers/douyineffect) | 抖音特效 | 30 | | characterize | [click](https://mp.weixin.qq.com/s/yaNQJyeUeisOenEeoVsgDg) | N/A | [click](./pydrawing/modules/beautifiers/characterize) | 视频转字符画 | 31 | | cartoonise | [click](https://mp.weixin.qq.com/s/efwNQl0JVJt6_x_evdL41A) | N/A | [click](./pydrawing/modules/beautifiers/cartoonise) | 图像卡通化 | 32 | | photomosaic | [click](https://mp.weixin.qq.com/s/BG1VW3jx0LUazhhifBapVw) | N/A | [click](./pydrawing/modules/beautifiers/photomosaic) | 拼马赛克图片 | 33 | | beziercurve | [click](https://mp.weixin.qq.com/s/SWpaTPw9tOLs5h1EgP30Vw) | N/A | [click](./pydrawing/modules/beautifiers/beziercurve) | 贝塞尔曲线画画 | 34 | | geneticfittingcircle | [click](https://mp.weixin.qq.com/s/L0z1ZO1Qztk0EF1KAMfmbA) | N/A | [click](./pydrawing/modules/beautifiers/geneticfitting) | 遗传算法拟合图像-圆形 | 35 | | geneticfittingpolygon | [click](https://mp.weixin.qq.com/s/L0z1ZO1Qztk0EF1KAMfmbA) | N/A | [click](./pydrawing/modules/beautifiers/geneticfitting) | 遗传算法拟合图像-多边形 | 36 | | nostalgicstyle | [click](https://mp.weixin.qq.com/s/yRCt69u_gzPI85-vOrb_sQ) | N/A | [click](./pydrawing/modules/beautifiers/nostalgicstyle) | 照片怀旧风格 | 37 | | photocorrection | [click](https://mp.weixin.qq.com/s/yRCt69u_gzPI85-vOrb_sQ) | N/A | [click](./pydrawing/modules/beautifiers/photocorrection) | 简单的照片矫正 | 38 | | pencildrawing | [click](https://mp.weixin.qq.com/s/K_2lGGlLKHIIm4iSg0xCUw) | [click](https://jiaya.me/archive/projects/pencilsketch/npar12_pencil.pdf) | [click](./pydrawing/modules/beautifiers/pencildrawing) | 铅笔素描画 | 39 | | cartoongan | [click](https://mp.weixin.qq.com/s/18fUOO5fH1PVUzTMNNCWwQ) | [click](https://openaccess.thecvf.com/content_cvpr_2018/CameraReady/2205.pdf) | [click](./pydrawing/modules/beautifiers/cartoongan) | 卡通GAN | 40 | | fastneuralstyletransfer | [click](https://mp.weixin.qq.com/s/Ed-1fWOIhI52G-Ugrv7n9Q) | [click](https://cs.stanford.edu/people/jcjohns/papers/eccv16/JohnsonECCV16.pdf) | [click](./pydrawing/modules/beautifiers/fastneuralstyletransfer) | 快速风格迁移 | 41 | | cartoonizeface | [click](https://mp.weixin.qq.com/s/L0z1ZO1Qztk0EF1KAMfmbA) | [click](https://arxiv.org/pdf/1907.10830.pdf) | [click](./pydrawing/modules/beautifiers/cartoonizeface) | 人脸卡通化 | 42 | | noteprocessor | [click](https://mp.weixin.qq.com/s/yRCt69u_gzPI85-vOrb_sQ) | [click](https://mzucker.github.io/2016/09/20/noteshrink.html) | [click](./pydrawing/modules/beautifiers/noteprocessor) | 手写笔记处理 | 43 | | oilpainting | [click](https://mp.weixin.qq.com/s/yRCt69u_gzPI85-vOrb_sQ) | [click](https://github.com/cyshih73/Faster-OilPainting/blob/master/Report.pdf) | [click](./pydrawing/modules/beautifiers/oilpainting) | 照片油画化 | 44 | 45 | 46 | # Install 47 | 48 | #### Preparation 49 | - [ffmpeg](https://ffmpeg.org/): You should set ffmpeg in environment variable. 50 | - [Pytorch](https://pytorch.org/get-started/previous-versions/): To apply some of the supported beautifiers (e.g., cartoongan), you need to install pytorch and corresponding dependencies following [tutorial](https://pytorch.org/get-started/previous-versions/). 51 | 52 | #### Pip install 53 | ```sh 54 | run "pip install pydrawing" 55 | ``` 56 | 57 | #### Source code install 58 | ```sh 59 | (1) Offline 60 | Step1: git clone https://github.com/CharlesPikachu/pydrawing.git 61 | Step2: cd pydrawing -> run "python setup.py install" 62 | (2) Online 63 | run "pip install git+https://github.com/CharlesPikachu/pydrawing.git@master" 64 | ``` 65 | 66 | 67 | # Quick Start 68 | ```python 69 | import random 70 | from pydrawing import pydrawing 71 | 72 | filepath = 'asserts/dog.jpg' 73 | config = { 74 | "savedir": "outputs", 75 | "savename": "output" 76 | } 77 | drawing_client = pydrawing.pydrawing() 78 | drawing_client.execute(filepath, random.choice(drawing_client.getallsupports()), config=config) 79 | ``` 80 | 81 | 82 | # Screenshot 83 |
84 | 85 |
86 |
87 |
88 | 89 |
90 |
91 |
92 | 93 |
94 |
95 |
96 | 97 |
98 |
99 | 100 | 101 | # Projects in Charles_pikachu 102 | - [Games](https://github.com/CharlesPikachu/Games): Create interesting games by pure python. 103 | - [DecryptLogin](https://github.com/CharlesPikachu/DecryptLogin): APIs for loginning some websites by using requests. 104 | - [Musicdl](https://github.com/CharlesPikachu/musicdl): A lightweight music downloader written by pure python. 105 | - [Videodl](https://github.com/CharlesPikachu/videodl): A lightweight video downloader written by pure python. 106 | - [Pytools](https://github.com/CharlesPikachu/pytools): Some useful tools written by pure python. 107 | - [PikachuWeChat](https://github.com/CharlesPikachu/pikachuwechat): Play WeChat with itchat-uos. 108 | - [Pydrawing](https://github.com/CharlesPikachu/pydrawing): Beautify your image or video. 109 | - [ImageCompressor](https://github.com/CharlesPikachu/imagecompressor): Image compressors written by pure python. 110 | - [FreeProxy](https://github.com/CharlesPikachu/freeproxy): Collecting free proxies from internet. 111 | - [Paperdl](https://github.com/CharlesPikachu/paperdl): Search and download paper from specific websites. 112 | - [Sciogovterminal](https://github.com/CharlesPikachu/sciogovterminal): Browse "The State Council Information Office of the People's Republic of China" in the terminal. 113 | - [CodeFree](https://github.com/CharlesPikachu/codefree): Make no code a reality. 114 | - [DeepLearningToys](https://github.com/CharlesPikachu/deeplearningtoys): Some deep learning toys implemented in pytorch. 115 | - [DataAnalysis](https://github.com/CharlesPikachu/dataanalysis): Some data analysis projects in charles_pikachu. 116 | - [Imagedl](https://github.com/CharlesPikachu/imagedl): Search and download images from specific websites. 117 | - [Pytoydl](https://github.com/CharlesPikachu/pytoydl): A toy deep learning framework built upon numpy. 118 | - [NovelDL](https://github.com/CharlesPikachu/noveldl): Search and download novels from some specific websites. 119 | 120 | 121 | # More 122 | #### WeChat Official Accounts 123 | *Charles_pikachu* 124 | ![img](./docs/pikachu.jpg) -------------------------------------------------------------------------------- /docs/Author.md: -------------------------------------------------------------------------------- 1 | # 关于作者 2 | 3 | 学生党, 主要研究方向是计算机视觉, 顺便对信息安全感兴趣。 4 | 5 | 我的个人微信公众号是: Charles_pikachu (欢迎搜索关注，或者搜"Charles的皮卡丘") 6 | 7 | 我的Github账号是: [https://github.com/CharlesPikachu](https://github.com/CharlesPikachu) (欢迎搜索关注) 8 | 9 | 我的知乎账号是: [https://www.zhihu.com/people/charles_pikachu](https://www.zhihu.com/people/charles_pikachu) (欢迎搜索关注) 10 | 11 | 我的B站账号是: [https://space.bilibili.com/406756145](https://space.bilibili.com/406756145) (欢迎搜索关注) 12 | 13 | 个人邮箱: charlesblwx@gmail.com -------------------------------------------------------------------------------- /docs/Changelog.md: -------------------------------------------------------------------------------- 1 | # 开发日志 2 | 3 | **2022-01-18** 4 | 5 | - 版本号: v0.1.0, 6 | - 更新内容: 支持多种图像/视频美化算法。 7 | 8 | **2022-01-19** 9 | 10 | - 版本号: v0.1.1到v0.1.2, 11 | - 更新内容: 部分算法优化, 支持人脸卡通化算法。 12 | 13 | **2022-01-25** 14 | 15 | - 版本号: v0.1.3到v0.1.4, 16 | - 更新内容: 支持遗传算法拟合图像和贝塞尔曲线画画, 对需要pytorch的算法进行优化, 避免不安装pytorch则无法使用其他功能的尴尬以及解决不安装cuda则无法导入模型的问题。 17 | 18 | **2022-01-26** 19 | 20 | - 版本号: v0.1.5, 21 | - 更新内容: 修复输入输出视频FPS不一致的问题。 22 | 23 | **2022-02-27** 24 | 25 | - 版本号: v0.1.6, 26 | - 更新内容: 支持手写笔记处理、照片怀旧风格等算法。 27 | 28 | **2022-03-21** 29 | 30 | - 版本号: v0.1.7, 31 | - 更新内容: 修改开源协议到Apache-2.0。 32 | 33 | **2022-03-24** 34 | 35 | - 版本号: v0.1.8, 36 | - 更新内容: 增加作者信息。 37 | 38 | **2022-04-23** 39 | 40 | - 版本号: v0.1.9, 41 | - 更新内容: 拼马赛克图片支持多种图片类型。 -------------------------------------------------------------------------------- /docs/Install.md: -------------------------------------------------------------------------------- 1 | # 安装Pydrawing 2 | 3 | 4 | #### 环境配置 5 | 6 | - 操作系统: Linux or macOS or Windows 7 | - Python版本: Python3.6+ 8 | - ffmpeg: 若输入视频中含有音频, 需要借助[ffmpeg](https://ffmpeg.org/)解码, 因此需要保证电脑中存在ffmpeg并在环境变量中。 9 | - Pytorch: 若需要使用CartoonGan等算法, 需要安装Pytorch>=1.0.0和配置对应的环境, 详见[官方文档](https://pytorch.org/get-started/locally/)。 10 | 11 | 12 | #### PIP安装(推荐) 13 | 14 | 在终端运行如下命令即可(请保证python在环境变量中): 15 | 16 | ```sh 17 | pip install pydrawing --upgrade 18 | ``` 19 | 20 | 21 | #### 源代码安装 22 | 23 | **1.在线安装** 24 | 25 | 运行如下命令即可在线安装: 26 | 27 | ```sh 28 | pip install git+https://github.com/CharlesPikachu/pydrawing.git@master 29 | ``` 30 | 31 | **2.离线安装** 32 | 33 | 利用如下命令下载pydrawing源代码到本地: 34 | 35 | ```sh 36 | git clone https://github.com/CharlesPikachu/pydrawing.git 37 | ``` 38 | 39 | 接着, 切到pydrawing目录下: 40 | 41 | ```sh 42 | cd pydrawing 43 | ``` 44 | 45 | 最后运行如下命令进行安装: 46 | 47 | ```sh 48 | python setup.py install 49 | ``` -------------------------------------------------------------------------------- /docs/Makefile: -------------------------------------------------------------------------------- 1 | # Minimal makefile for Sphinx documentation 2 | # 3 | 4 | # You can set these variables from the command line, and also 5 | # from the environment for the first two. 6 | SPHINXOPTS ?= 7 | SPHINXBUILD ?= sphinx-build 8 | SOURCEDIR = source 9 | BUILDDIR = build 10 | 11 | # Put it first so that "make" without argument is like "make help". 12 | help: 13 | @$(SPHINXBUILD) -M help "$(SOURCEDIR)" "$(BUILDDIR)" $(SPHINXOPTS) $(O) 14 | 15 | .PHONY: help Makefile 16 | 17 | # Catch-all target: route all unknown targets to Sphinx using the new 18 | # "make mode" option. $(O) is meant as a shortcut for $(SPHINXOPTS). 19 | %: Makefile 20 | @$(SPHINXBUILD) -M $@ "$(SOURCEDIR)" "$(BUILDDIR)" $(SPHINXOPTS) $(O) 21 | -------------------------------------------------------------------------------- /docs/Quickstart.md: -------------------------------------------------------------------------------- 1 | # 快速开始 2 | 3 | 4 | ## 已经支持的算法 5 | 6 | #### 图像卡通化 7 | 8 | **1.相关论文** 9 | 10 | 暂无 11 | 12 | **2.公众号文章介绍** 13 | 14 | [Introduction](https://mp.weixin.qq.com/s/efwNQl0JVJt6_x_evdL41A) 15 | 16 | **3.调用示例** 17 | 18 | ```python 19 | from pydrawing import pydrawing 20 | 21 | config = {'mode': ['rgb', 'hsv'][0]} 22 | filepath = 'input.jpg' 23 | drawing_client = pydrawing.pydrawing() 24 | drawing_client.execute(filepath, 'cartoonise', config=config) 25 | ``` 26 | 27 | **4.config选项** 28 | 29 | - savename: 保存结果时用的文件名, 默认值为"output"; 30 | - savedir: 保存结果时用的文件夹, 默认值为"outputs"; 31 | - merge_audio: 处理视频时, 是否把原视频中的音频合成到生成的视频中, 默认值为"False"; 32 | - mode: 卡通化时所用的颜色空间, 支持"rgb"和"hsv"模式, 默认值为"rgb"。 33 | 34 | #### 人脸卡通化 35 | 36 | **1.相关论文** 37 | 38 | [Paper](https://arxiv.org/pdf/1907.10830.pdf) 39 | 40 | **2.公众号文章介绍** 41 | 42 | [Introduction](https://mp.weixin.qq.com/s/L0z1ZO1Qztk0EF1KAMfmbA) 43 | 44 | **3.调用示例** 45 | 46 | ```python 47 | from pydrawing import pydrawing 48 | 49 | config = {'use_face_segmentor': False} 50 | filepath = 'input.jpg' 51 | drawing_client = pydrawing.pydrawing() 52 | drawing_client.execute(filepath, 'cartoonizeface', config=config) 53 | ``` 54 | 55 | **4.config选项** 56 | 57 | - savename: 保存结果时用的文件名, 默认值为"output"; 58 | - savedir: 保存结果时用的文件夹, 默认值为"outputs"; 59 | - merge_audio: 处理视频时, 是否把原视频中的音频合成到生成的视频中, 默认值为"False"; 60 | - use_cuda: 模型是否使用cuda加速, 默认值为"False"; 61 | - use_face_segmentor: 是否使用人脸分割器进一步去除人脸背景, 默认值为"False"。 62 | 63 | #### 铅笔素描画 64 | 65 | **1.相关论文** 66 | 67 | [Paper](https://jiaya.me/archive/projects/pencilsketch/npar12_pencil.pdf) 68 | 69 | **2.公众号文章介绍** 70 | 71 | [Introduction](https://mp.weixin.qq.com/s/K_2lGGlLKHIIm4iSg0xCUw) 72 | 73 | **3.调用示例** 74 | 75 | ```python 76 | from pydrawing import pydrawing 77 | 78 | config = {'mode': ['gray', 'color'][0]} 79 | filepath = 'input.jpg' 80 | drawing_client = pydrawing.pydrawing() 81 | drawing_client.execute(filepath, 'pencildrawing', config=config) 82 | ``` 83 | 84 | **4.config选项** 85 | 86 | - savename: 保存结果时用的文件名, 默认值为"output"; 87 | - savedir: 保存结果时用的文件夹, 默认值为"outputs"; 88 | - merge_audio: 处理视频时, 是否把原视频中的音频合成到生成的视频中, 默认值为"False"; 89 | - mode: 生成的图片是灰色图还是彩色图, 支持"gray"和"color", 默认值为"gray"; 90 | - kernel_size_scale: 铅笔笔画相关参数, 默认值为"1/40"; 91 | - stroke_width: 铅笔笔画相关参数, 默认值为"1"; 92 | - color_depth: 铅笔色调相关参数, 默认值为"1"; 93 | - weights_color: 铅笔色调相关参数, 默认值为"[62, 30, 5]"; 94 | - weights_gray: 铅笔色调相关参数, 默认值为"[76, 22, 2]"; 95 | - texture_path: 纹理图片路径, 默认使用库里提供的"default.jpg"文件。 96 | 97 | #### 卡通GAN 98 | 99 | **1.相关论文** 100 | 101 | [Paper](https://openaccess.thecvf.com/content_cvpr_2018/CameraReady/2205.pdf) 102 | 103 | **2.公众号文章介绍** 104 | 105 | [Introduction](https://mp.weixin.qq.com/s/18fUOO5fH1PVUzTMNNCWwQ) 106 | 107 | **3.调用示例** 108 | ```python 109 | from pydrawing import pydrawing 110 | 111 | config = {'style': ['Hayao', 'Hosoda', 'Paprika', 'Shinkai'][0]} 112 | filepath = 'input.jpg' 113 | drawing_client = pydrawing.pydrawing() 114 | drawing_client.execute(filepath, 'cartoongan', config=config) 115 | ``` 116 | 117 | **4.config选项** 118 | 119 | - savename: 保存结果时用的文件名, 默认值为"output"; 120 | - savedir: 保存结果时用的文件夹, 默认值为"outputs"; 121 | - merge_audio: 处理视频时, 是否把原视频中的音频合成到生成的视频中, 默认值为"False"; 122 | - style: 卡通画的风格类型, 支持"Hayao", "Hosoda", "Paprika"和"Shinkai", 默认值为"Hosoda"; 123 | - use_cuda: 模型是否使用cuda加速, 默认值为"True"。 124 | 125 | #### 快速风格迁移 126 | 127 | **1.相关论文** 128 | 129 | [Paper](https://cs.stanford.edu/people/jcjohns/papers/eccv16/JohnsonECCV16.pdf) 130 | 131 | **2.公众号文章介绍** 132 | 133 | [Introduction](https://mp.weixin.qq.com/s/Ed-1fWOIhI52G-Ugrv7n9Q) 134 | 135 | **3.调用示例** 136 | 137 | ```python 138 | from pydrawing import pydrawing 139 | 140 | config = {'style': ['starrynight', 'cuphead', 'mosaic​'][0]} 141 | filepath = 'input.jpg' 142 | drawing_client = pydrawing() 143 | drawing_client.execute(filepath, 'fastneuralstyletransfer', config=config) 144 | ``` 145 | 146 | **4.config选项** 147 | 148 | - savename: 保存结果时用的文件名, 默认值为"output"; 149 | - savedir: 保存结果时用的文件夹, 默认值为"outputs"; 150 | - merge_audio: 处理视频时, 是否把原视频中的音频合成到生成的视频中, 默认值为"False"; 151 | - style: 迁移的画的风格类型, 支持"starrynight", "cuphead"和"mosaic", 默认值为"starrynight"; 152 | - use_cuda: 模型是否使用cuda加速, 默认值为"True"。 153 | 154 | #### 抖音特效 155 | 156 | **1.相关论文** 157 | 158 | 暂无 159 | 160 | **2.公众号文章介绍** 161 | 162 | [Introduction](https://mp.weixin.qq.com/s/RRnrO2H84pvtUdDsAYD9Qg) 163 | 164 | **3.调用示例** 165 | 166 | ```python 167 | from pydrawing import pydrawing 168 | 169 | filepath = 'input.jpg' 170 | drawing_client = pydrawing.pydrawing() 171 | drawing_client.execute(filepath, 'douyineffect') 172 | ``` 173 | 174 | **4.config选项** 175 | 176 | - savename: 保存结果时用的文件名, 默认值为"output"; 177 | - savedir: 保存结果时用的文件夹, 默认值为"outputs"; 178 | - merge_audio: 处理视频时, 是否把原视频中的音频合成到生成的视频中, 默认值为"False"。 179 | 180 | #### 视频转字符画 181 | 182 | **1.相关论文** 183 | 184 | 暂无 185 | 186 | **2.公众号文章介绍** 187 | 188 | [Introduction](https://mp.weixin.qq.com/s/yaNQJyeUeisOenEeoVsgDg) 189 | 190 | **3.调用示例** 191 | 192 | ```python 193 | from pydrawing import pydrawing 194 | 195 | filepath = 'input.jpg' 196 | drawing_client = pydrawing.pydrawing() 197 | drawing_client.execute(filepath, 'characterize') 198 | ``` 199 | 200 | **4.config选项** 201 | 202 | - savename: 保存结果时用的文件名, 默认值为"output"; 203 | - savedir: 保存结果时用的文件夹, 默认值为"outputs"; 204 | - merge_audio: 处理视频时, 是否把原视频中的音频合成到生成的视频中, 默认值为"False"。 205 | 206 | #### 拼马赛克图片 207 | 208 | **1.相关论文** 209 | 210 | 暂无 211 | 212 | **2.公众号文章介绍** 213 | 214 | [Introduction](https://mp.weixin.qq.com/s/BG1VW3jx0LUazhhifBapVw) 215 | 216 | **3.调用示例** 217 | 218 | ```python 219 | from pydrawing import pydrawing 220 | ​ 221 | config = {'src_images_dir': 'images', 'block_size': 15} 222 | filepath = 'input.jpg' 223 | drawing_client = pydrawing.pydrawing() 224 | drawing_client.execute(filepath, 'photomosaic', config=config) 225 | ``` 226 | 227 | **4.config选项** 228 | 229 | - savename: 保存结果时用的文件名, 默认值为"output"; 230 | - savedir: 保存结果时用的文件夹, 默认值为"outputs"; 231 | - merge_audio: 处理视频时, 是否把原视频中的音频合成到生成的视频中, 默认值为"False"; 232 | - block_size: 马赛克block大小, 默认值为"15"; 233 | - src_images_dir: 使用的图片路径, 请保证该文件夹中存在大量色彩各异的图片以实现较好的拼图效果。 234 | 235 | #### 信号故障特效 236 | 237 | **1.相关论文** 238 | 239 | 暂无 240 | 241 | **2.公众号文章介绍** 242 | 243 | [Introduction](https://mp.weixin.qq.com/s/Yv0uPLsTGwVnj_PKqYCmAw) 244 | 245 | **3.调用示例** 246 | 247 | ```python 248 | from pydrawing import pydrawing 249 | 250 | filepath = 'input.mp4' 251 | drawing_client = pydrawing.pydrawing() 252 | drawing_client.execute(filepath, 'glitch') 253 | ``` 254 | 255 | **4.config选项** 256 | 257 | - savename: 保存结果时用的文件名, 默认值为"output"; 258 | - savedir: 保存结果时用的文件夹, 默认值为"outputs"; 259 | - merge_audio: 处理视频时, 是否把原视频中的音频合成到生成的视频中, 默认值为"False"; 260 | - header_size: 文件头部大小, 一般不需要改, 默认值为"200"; 261 | - intensity: 随机扰动相关的参数, 默认值为"0.1"; 262 | - block_size: 一次读取文件的大小, 默认值为"100"。 263 | 264 | #### 贝塞尔曲线画画 265 | 266 | **1.相关论文** 267 | 268 | 暂无 269 | 270 | **2.公众号文章介绍** 271 | 272 | [Introduction](https://mp.weixin.qq.com/s/SWpaTPw9tOLs5h1EgP30Vw) 273 | 274 | **3.调用示例** 275 | 276 | ```python 277 | from pydrawing import pydrawing 278 | 279 | filepath = 'input.jpg' 280 | drawing_client = pydrawing.pydrawing() 281 | drawing_client.execute(filepath, 'beziercurve') 282 | ``` 283 | 284 | **4.config选项** 285 | 286 | - savename: 保存结果时用的文件名, 默认值为"output"; 287 | - savedir: 保存结果时用的文件夹, 默认值为"outputs"; 288 | - merge_audio: 处理视频时, 是否把原视频中的音频合成到生成的视频中, 默认值为"False"; 289 | - num_samples: 采样点, 默认值为"15"; 290 | - width: 坐标变换宽度, 默认值为"600"; 291 | - height: 坐标变换高度, 默认值为"600"; 292 | - num_colors: 使用的颜色数量, 默认值为"32"。 293 | 294 | #### 遗传算法拟合图像-圆形 295 | 296 | **1.相关论文** 297 | 298 | 暂无 299 | 300 | **2.公众号文章介绍** 301 | 302 | [Introduction](https://mp.weixin.qq.com/s/L0z1ZO1Qztk0EF1KAMfmbA) 303 | 304 | **3.调用示例** 305 | 306 | ```python 307 | from pydrawing import pydrawing 308 | 309 | filepath = 'input.jpg' 310 | drawing_client = pydrawing.pydrawing() 311 | drawing_client.execute(filepath, 'geneticfittingcircle') 312 | ``` 313 | 314 | **4.config选项** 315 | 316 | - savename: 保存结果时用的文件名, 默认值为"output"; 317 | - savedir: 保存结果时用的文件夹, 默认值为"outputs"; 318 | - merge_audio: 处理视频时, 是否把原视频中的音频合成到生成的视频中, 默认值为"False"; 319 | - cache_dir: 中间结果保存的文件夹, 默认值为"cache"; 320 | - save_cache: 是否保存中间结果, 默认值为"True"; 321 | - init_cfg: 算法初始化参数, 默认值为如下: 322 | ```python 323 | init_cfg = { 324 | 'num_populations': 10, 325 | 'init_num_circles': 1, 326 | 'num_generations': 1e5, 327 | 'print_interval': 1, 328 | 'mutation_rate': 0.1, 329 | 'selection_rate': 0.5, 330 | 'crossover_rate': 0.5, 331 | 'circle_cfg': {'radius_range': 50, 'radius_shift_range': 50, 'center_shift_range': 50, 'color_shift_range': 50}, 332 | } 333 | ``` 334 | 335 | #### 遗传算法拟合图像-多边形 336 | 337 | **1.相关论文** 338 | 339 | 暂无 340 | 341 | **2.公众号文章介绍** 342 | 343 | [Introduction](https://mp.weixin.qq.com/s/L0z1ZO1Qztk0EF1KAMfmbA) 344 | 345 | **3.调用示例** 346 | 347 | ```python 348 | from pydrawing import pydrawing 349 | 350 | filepath = 'input.jpg' 351 | drawing_client = pydrawing.pydrawing() 352 | drawing_client.execute(filepath, 'geneticfittingpolygon') 353 | ``` 354 | 355 | **4.config选项** 356 | 357 | - savename: 保存结果时用的文件名, 默认值为"output"; 358 | - savedir: 保存结果时用的文件夹, 默认值为"outputs"; 359 | - merge_audio: 处理视频时, 是否把原视频中的音频合成到生成的视频中, 默认值为"False"; 360 | - cache_dir: 中间结果保存的文件夹, 默认值为"cache"; 361 | - save_cache: 是否保存中间结果, 默认值为"True"; 362 | - init_cfg: 算法初始化参数, 默认值为如下: 363 | ```python 364 | init_cfg = { 365 | 'num_populations': 10, 366 | 'num_points_list': list(range(3, 40)), 367 | 'init_num_polygons': 1, 368 | 'num_generations': 1e5, 369 | 'print_interval': 1, 370 | 'mutation_rate': 0.1, 371 | 'selection_rate': 0.5, 372 | 'crossover_rate': 0.5, 373 | 'polygon_cfg': {'size': 50, 'shift_range': 50, 'point_range': 50, 'color_range': 50}, 374 | } 375 | ``` 376 | 377 | #### 照片怀旧风格 378 | 379 | **1.相关论文** 380 | 381 | 暂无 382 | 383 | **2.公众号文章介绍** 384 | 385 | [Introduction](https://mp.weixin.qq.com/s/yRCt69u_gzPI85-vOrb_sQ) 386 | 387 | **3.调用示例** 388 | 389 | ```python 390 | from pydrawing import pydrawing 391 | ​ 392 | filepath = 'input.jpg' 393 | drawing_client = pydrawing() 394 | drawing_client.execute(filepath, 'nostalgicstyle') 395 | ``` 396 | 397 | **4.config选项** 398 | 399 | - savename: 保存结果时用的文件名, 默认值为"output"; 400 | - savedir: 保存结果时用的文件夹, 默认值为"outputs"; 401 | - merge_audio: 处理视频时, 是否把原视频中的音频合成到生成的视频中, 默认值为"False"。 402 | 403 | #### 手写笔记处理 404 | 405 | **1.相关论文** 406 | 407 | [Paper](https://mzucker.github.io/2016/09/20/noteshrink.html) 408 | 409 | **2.公众号文章介绍** 410 | 411 | [Introduction](https://mp.weixin.qq.com/s/yRCt69u_gzPI85-vOrb_sQ) 412 | 413 | **3.调用示例** 414 | 415 | ```python 416 | from pydrawing import pydrawing 417 | ​ 418 | config = { 419 | 'sat_threshold': 0.20, 420 | 'value_threshold': 0.25, 421 | 'num_colors': 8, 422 | 'sample_fraction': 0.05, 423 | 'white_bg': False, 424 | 'saturate': True, 425 | } 426 | filepath = 'input.jpg' 427 | drawing_client = pydrawing() 428 | drawing_client.execute(filepath, 'noteprocessor', config=config) 429 | ``` 430 | 431 | **4.config选项** 432 | 433 | - savename: 保存结果时用的文件名, 默认值为"output"; 434 | - savedir: 保存结果时用的文件夹, 默认值为"outputs"; 435 | - merge_audio: 处理视频时, 是否把原视频中的音频合成到生成的视频中, 默认值为"False"; 436 | - sat_threshold: 背景饱和度阈值, 默认值为"0.2"; 437 | - value_threshold: 背景的阈值, 默认值为"0.25"; 438 | - num_colors: 输出颜色的数量, 默认值为"8"; 439 | - sample_fraction: 采样的像素占比, 默认值为"0.05"; 440 | - white_bg: 使背景为白色, 默认值为"False"; 441 | - saturate: 使颜色不饱和, 默认值为"True"。 442 | 443 | #### 照片油画化 444 | 445 | **1.相关论文** 446 | 447 | [Paper](https://github.com/cyshih73/Faster-OilPainting/blob/master/Report.pdf) 448 | 449 | **2.公众号文章介绍** 450 | 451 | [Introduction](https://mp.weixin.qq.com/s/yRCt69u_gzPI85-vOrb_sQ) 452 | 453 | **3.调用示例** 454 | 455 | ```python 456 | from pydrawing import pydrawing 457 | ​ 458 | config = { 459 | 'edge_operator': 'sobel', 460 | 'palette': 0, 461 | 'brush_width': 5, 462 | } 463 | filepath = 'input.jpg' 464 | drawing_client = pydrawing() 465 | drawing_client.execute(filepath, 'oilpainting​', config=config) 466 | ``` 467 | 468 | **4.config选项** 469 | 470 | - savename: 保存结果时用的文件名, 默认值为"output"; 471 | - savedir: 保存结果时用的文件夹, 默认值为"outputs"; 472 | - merge_audio: 处理视频时, 是否把原视频中的音频合成到生成的视频中, 默认值为"False"; 473 | - brush_width: 画笔大小, 默认值为"5"; 474 | - palette: 调色板颜色, 默认为"0", 代表使用原图的实际颜色; 475 | - edge_operator: 使用的边缘检测算子, 支持"sobel", "prewitt", "scharr"和"roberts", 默认值为"sobel"。 476 | 477 | #### 简单的照片矫正 478 | 479 | **1.相关论文** 480 | 481 | 暂无。 482 | 483 | **2.公众号文章介绍** 484 | 485 | [Introduction](https://mp.weixin.qq.com/s/yRCt69u_gzPI85-vOrb_sQ) 486 | 487 | **3.调用示例** 488 | 489 | ```python 490 | from pydrawing import pydrawing 491 | ​ 492 | config = { 493 | 'epsilon_factor': 0.08, 494 | 'canny_boundaries': [100, 200], 495 | 'use_preprocess': False, 496 | } 497 | filepath = 'input.jpg' 498 | drawing_client = pydrawing() 499 | drawing_client.execute(filepath, 'photocorrection', config=config) 500 | ``` 501 | 502 | **4.config选项** 503 | 504 | - savename: 保存结果时用的文件名, 默认值为"output"; 505 | - savedir: 保存结果时用的文件夹, 默认值为"outputs"; 506 | - merge_audio: 处理视频时, 是否把原视频中的音频合成到生成的视频中, 默认值为"False"; 507 | - epsilon_factor: 多边形估计时的超参数, 默认为"0.08"; 508 | - canny_boundaries: canny边缘检测算子的两个边界值, 默认为"[100, 200]"; 509 | - use_preprocess: 是否在边缘检测前对图像进行预处理, 默认值为"False"。 510 | 511 | 512 | ## 随机运行一个小程序 513 | 514 | 写如下代码，保存并运行即可: 515 | 516 | ```python 517 | import random 518 | from pydrawing import pydrawing 519 | 520 | filepath = 'asserts/dog.jpg' 521 | config = { 522 | "savedir": "outputs", 523 | "savename": "output" 524 | } 525 | drawing_client = pydrawing.pydrawing() 526 | drawing_client.execute(filepath, random.choice(drawing_client.getallsupports()), config=config) 527 | ``` 528 | 529 | 部分测试效果如下: 530 | 531 |
532 | 533 |
534 |
535 |
536 | 537 |
538 |
539 |
540 | 541 |
542 |
543 |
544 | 545 |
546 |
-------------------------------------------------------------------------------- /docs/Recommend.md: -------------------------------------------------------------------------------- 1 | # 项目推荐 2 | 3 | - [制作小游戏](https://github.com/CharlesPikachu/Games) 4 | 5 | - [模拟登录系列](https://github.com/CharlesPikachu/DecryptLogin) 6 | 7 | - [音乐下载器](https://github.com/CharlesPikachu/musicdl) 8 | 9 | - [视频下载器](https://github.com/CharlesPikachu/videodl) 10 | 11 | - [实用工具](https://github.com/CharlesPikachu/pytools) 12 | 13 | - [玩转微信](https://github.com/CharlesPikachu/pikachuwechat) 14 | 15 | - [图像语义分割框架](https://github.com/SegmentationBLWX/sssegmentation) 16 | 17 | - [美化图片或视频](https://github.com/CharlesPikachu/pydrawing) 18 | 19 | - [图像压缩算法](https://github.com/CharlesPikachu/imagecompressor) 20 | 21 | - [免费代理工具](https://github.com/CharlesPikachu/freeproxy) 22 | 23 | - [美丽的星空图](https://github.com/CharlesPikachu/constellation) 24 | 25 | - [论文下载器](https://github.com/CharlesPikachu/paperdl) 26 | 27 | - [终端看中华人民共和国国务院新闻办公室](https://github.com/CharlesPikachu/sciogovterminal) 28 | 29 | - [代码自由](https://github.com/CharlesPikachu/codefree) 30 | 31 | - [深度学习小案例](https://github.com/CharlesPikachu/deeplearningtoys) 32 | 33 | - [数据分析相关的小项目](https://github.com/CharlesPikachu/dataanalysis) 34 | 35 | - [图片下载器](https://github.com/CharlesPikachu/imagedl) 36 | 37 | - [从零开始实现一个深度学习框架](https://github.com/CharlesPikachu/pytoydl) 38 | 39 | - [小说下载器](https://github.com/CharlesPikachu/noveldl) 40 | -------------------------------------------------------------------------------- /docs/State.md: -------------------------------------------------------------------------------- 1 | # 项目声明 2 | 3 |
4 | 5 |
6 |
7 | 8 | 本项目仅供python爱好者学习使用, 禁止用于商业用途, 希望大家合理利用该项目🙂 -------------------------------------------------------------------------------- /docs/conf.py: -------------------------------------------------------------------------------- 1 | # Configuration file for the Sphinx documentation builder. 2 | # 3 | # This file only contains a selection of the most common options. For a full 4 | # list see the documentation: 5 | # https://www.sphinx-doc.org/en/master/usage/configuration.html 6 | 7 | # -- Path setup -------------------------------------------------------------- 8 | 9 | # If extensions (or modules to document with autodoc) are in another directory, 10 | # add these directories to sys.path here. If the directory is relative to the 11 | # documentation root, use os.path.abspath to make it absolute, like shown here. 12 | # 13 | # import os 14 | # import sys 15 | # sys.path.insert(0, os.path.abspath('.')) 16 | 17 | 18 | # -- Project information ----------------------------------------------------- 19 | 20 | project = 'pydrawing' 21 | copyright = '2022, Zhenchao Jin' 22 | author = 'Zhenchao Jin' 23 | release = '0.1.0' 24 | 25 | # -- General configuration --------------------------------------------------- 26 | 27 | # Add any Sphinx extension module names here, as strings. They can be 28 | # extensions coming with Sphinx (named 'sphinx.ext.*') or your custom 29 | # ones. 30 | extensions = [ 31 | 'sphinx.ext.autodoc', 32 | 'sphinx.ext.napoleon', 33 | 'sphinx.ext.viewcode', 34 | 'recommonmark', 35 | 'sphinx_markdown_tables', 36 | ] 37 | 38 | # Add any paths that contain templates here, relative to this directory. 39 | templates_path = ['_templates'] 40 | 41 | # The suffix(es) of source filenames. 42 | # You can specify multiple suffix as a list of string: 43 | # 44 | source_suffix = { 45 | '.rst': 'restructuredtext', 46 | '.md': 'markdown', 47 | } 48 | 49 | # The master toctree document. 50 | master_doc = 'index' 51 | 52 | # List of patterns, relative to source directory, that match files and 53 | # directories to ignore when looking for source files. 54 | # This pattern also affects html_static_path and html_extra_path. 55 | exclude_patterns = ['_build', 'Thumbs.db', '.DS_Store'] 56 | 57 | 58 | # -- Options for HTML output ------------------------------------------------- 59 | 60 | # The theme to use for HTML and HTML Help pages. See the documentation for 61 | # a list of builtin themes. 62 | # 63 | html_theme = 'sphinx_rtd_theme' 64 | 65 | # Add any paths that contain custom static files (such as style sheets) here, 66 | # relative to this directory. They are copied after the builtin static files, 67 | # so a file named "default.css" will overwrite the builtin "default.css". 68 | html_static_path = ['_static'] 69 | 70 | # For multi language 71 | # locale_dirs = ['locale/'] 72 | # gettext_compact = False -------------------------------------------------------------------------------- /docs/index.rst: -------------------------------------------------------------------------------- 1 | .. pydrawing documentation master file, created by 2 | sphinx-quickstart on Sat Feb 29 22:07:23 2020. 3 | You can adapt this file completely to your liking, but it should at least 4 | contain the root `toctree` directive. 5 | 6 | Pydrawing中文文档 7 | ======================================== 8 | 9 | .. toctree:: 10 | :maxdepth: 2 11 | 12 | State.md 13 | Install.md 14 | Quickstart.md 15 | Changelog.md 16 | Recommend.md 17 | Author.md -------------------------------------------------------------------------------- /docs/logo.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/CharlesPikachu/pydrawing/e980ad9bf4cece42ff40ed2bc7bed7a155eabb8e/docs/logo.png -------------------------------------------------------------------------------- /docs/make.bat: -------------------------------------------------------------------------------- 1 | @ECHO OFF 2 | 3 | pushd %~dp0 4 | 5 | REM Command file for Sphinx documentation 6 | 7 | if "%SPHINXBUILD%" == "" ( 8 | set SPHINXBUILD=sphinx-build 9 | ) 10 | set SOURCEDIR=source 11 | set BUILDDIR=build 12 | 13 | if "%1" == "" goto help 14 | 15 | %SPHINXBUILD% >NUL 2>NUL 16 | if errorlevel 9009 ( 17 | echo. 18 | echo.The 'sphinx-build' command was not found. Make sure you have Sphinx 19 | echo.installed, then set the SPHINXBUILD environment variable to point 20 | echo.to the full path of the 'sphinx-build' executable. Alternatively you 21 | echo.may add the Sphinx directory to PATH. 22 | echo. 23 | echo.If you don't have Sphinx installed, grab it from 24 | echo.http://sphinx-doc.org/ 25 | exit /b 1 26 | ) 27 | 28 | %SPHINXBUILD% -M %1 %SOURCEDIR% %BUILDDIR% %SPHINXOPTS% %O% 29 | goto end 30 | 31 | :help 32 | %SPHINXBUILD% -M help %SOURCEDIR% %BUILDDIR% %SPHINXOPTS% %O% 33 | 34 | :end 35 | popd 36 | -------------------------------------------------------------------------------- /docs/pikachu.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/CharlesPikachu/pydrawing/e980ad9bf4cece42ff40ed2bc7bed7a155eabb8e/docs/pikachu.jpg -------------------------------------------------------------------------------- /docs/requirements.txt: -------------------------------------------------------------------------------- 1 | recommonmark 2 | sphinx==4.5.0 3 | sphinx_markdown_tables==0.0.12 4 | sphinx_rtd_theme -------------------------------------------------------------------------------- /docs/screenshot_characterize.gif: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/CharlesPikachu/pydrawing/e980ad9bf4cece42ff40ed2bc7bed7a155eabb8e/docs/screenshot_characterize.gif -------------------------------------------------------------------------------- /docs/screenshot_fastneuralstyletransfer.gif: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/CharlesPikachu/pydrawing/e980ad9bf4cece42ff40ed2bc7bed7a155eabb8e/docs/screenshot_fastneuralstyletransfer.gif -------------------------------------------------------------------------------- /docs/screenshot_photomosaic.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/CharlesPikachu/pydrawing/e980ad9bf4cece42ff40ed2bc7bed7a155eabb8e/docs/screenshot_photomosaic.png -------------------------------------------------------------------------------- /docs/screeshot_noteprocessor.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/CharlesPikachu/pydrawing/e980ad9bf4cece42ff40ed2bc7bed7a155eabb8e/docs/screeshot_noteprocessor.png -------------------------------------------------------------------------------- /pydrawing/__init__.py: -------------------------------------------------------------------------------- 1 | '''title''' 2 | __title__ = 'pydrawing' 3 | '''description''' 4 | __description__ = 'Pydrawing: Beautify your image or video.' 5 | '''url''' 6 | __url__ = 'https://github.com/CharlesPikachu/pydrawing' 7 | '''version''' 8 | __version__ = '0.1.9' 9 | '''author''' 10 | __author__ = 'Zhenchao Jin' 11 | '''email''' 12 | __email__ = 'charlesblwx@gmail.com' 13 | '''license''' 14 | __license__ = 'Apache License 2.0' 15 | '''copyright''' 16 | __copyright__ = 'Copyright 2021-2022 Zhenchao Jin' -------------------------------------------------------------------------------- /pydrawing/asserts/animecharacter.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/CharlesPikachu/pydrawing/e980ad9bf4cece42ff40ed2bc7bed7a155eabb8e/pydrawing/asserts/animecharacter.jpg -------------------------------------------------------------------------------- /pydrawing/asserts/badapple.mp4: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/CharlesPikachu/pydrawing/e980ad9bf4cece42ff40ed2bc7bed7a155eabb8e/pydrawing/asserts/badapple.mp4 -------------------------------------------------------------------------------- /pydrawing/asserts/badapple_color.mp4: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/CharlesPikachu/pydrawing/e980ad9bf4cece42ff40ed2bc7bed7a155eabb8e/pydrawing/asserts/badapple_color.mp4 -------------------------------------------------------------------------------- /pydrawing/asserts/bridge.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/CharlesPikachu/pydrawing/e980ad9bf4cece42ff40ed2bc7bed7a155eabb8e/pydrawing/asserts/bridge.png -------------------------------------------------------------------------------- /pydrawing/asserts/dog.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/CharlesPikachu/pydrawing/e980ad9bf4cece42ff40ed2bc7bed7a155eabb8e/pydrawing/asserts/dog.jpg -------------------------------------------------------------------------------- /pydrawing/asserts/face.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/CharlesPikachu/pydrawing/e980ad9bf4cece42ff40ed2bc7bed7a155eabb8e/pydrawing/asserts/face.jpg -------------------------------------------------------------------------------- /pydrawing/asserts/monalisa.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/CharlesPikachu/pydrawing/e980ad9bf4cece42ff40ed2bc7bed7a155eabb8e/pydrawing/asserts/monalisa.jpg -------------------------------------------------------------------------------- /pydrawing/asserts/note.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/CharlesPikachu/pydrawing/e980ad9bf4cece42ff40ed2bc7bed7a155eabb8e/pydrawing/asserts/note.jpg -------------------------------------------------------------------------------- /pydrawing/asserts/zurich.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/CharlesPikachu/pydrawing/e980ad9bf4cece42ff40ed2bc7bed7a155eabb8e/pydrawing/asserts/zurich.jpg -------------------------------------------------------------------------------- /pydrawing/modules/__init__.py: -------------------------------------------------------------------------------- 1 | '''initialize''' 2 | from .utils import Logger 3 | from .beautifiers import ( 4 | CartooniseBeautifier, PencilDrawingBeautifier, CartoonGanBeautifier, FastNeuralStyleTransferBeautifier, 5 | DouyinEffectBeautifier, CharacterizeBeautifier, PhotomosaicBeautifier, GlitchBeautifier, CartoonizeFaceBeautifier, 6 | GeneticFittingCircleBeautifier, GeneticFittingPolygonBeautifier, BezierCurveBeautifier, NostalgicstyleBeautifier, 7 | NoteprocessorBeautifier, OilpaintingBeautifier, PhotocorrectionBeautifier 8 | ) -------------------------------------------------------------------------------- /pydrawing/modules/beautifiers/__init__.py: -------------------------------------------------------------------------------- 1 | '''initialize''' 2 | from .glitch import GlitchBeautifier 3 | from .cartoonise import CartooniseBeautifier 4 | from .cartoongan import CartoonGanBeautifier 5 | from .oilpainting import OilpaintingBeautifier 6 | from .beziercurve import BezierCurveBeautifier 7 | from .photomosaic import PhotomosaicBeautifier 8 | from .characterize import CharacterizeBeautifier 9 | from .douyineffect import DouyinEffectBeautifier 10 | from .noteprocessor import NoteprocessorBeautifier 11 | from .pencildrawing import PencilDrawingBeautifier 12 | from .cartoonizeface import CartoonizeFaceBeautifier 13 | from .nostalgicstyle import NostalgicstyleBeautifier 14 | from .photocorrection import PhotocorrectionBeautifier 15 | from .fastneuralstyletransfer import FastNeuralStyleTransferBeautifier 16 | from .geneticfitting import GeneticFittingCircleBeautifier, GeneticFittingPolygonBeautifier -------------------------------------------------------------------------------- /pydrawing/modules/beautifiers/base/__init__.py: -------------------------------------------------------------------------------- 1 | '''initialize''' 2 | from .base import BaseBeautifier -------------------------------------------------------------------------------- /pydrawing/modules/beautifiers/base/base.py: -------------------------------------------------------------------------------- 1 | ''' 2 | Function: 3 | Beautifier基类 4 | Author: 5 | Charles 6 | 微信公众号: 7 | Charles的皮卡丘 8 | ''' 9 | import os 10 | import cv2 11 | import subprocess 12 | from tqdm import tqdm 13 | from ...utils import Images2VideoAndSave, SaveImage, ReadVideo 14 | 15 | 16 | '''Beautifier基类''' 17 | class BaseBeautifier(): 18 | def __init__(self, savedir='outputs', savename='output', **kwargs): 19 | self.savename, self.savedir = savename, savedir 20 | self.merge_audio, self.tmp_audio_path = False, 'cache.mp3' 21 | for key, value in kwargs.items(): setattr(self, key, value) 22 | '''处理文件''' 23 | def process(self, filepath, images=None): 24 | assert images is None or filepath is None, 'please input filepath or images rather than both' 25 | # 图片/视频处理 26 | if images is None: 27 | # --图片 28 | if filepath.split('.')[-1].lower() in ['jpg', 'jpeg', 'png']: 29 | images = [cv2.imread(filepath)] 30 | # --视频 31 | elif filepath.split('.')[-1].lower() in ['mp4', 'avi']: 32 | images, self.fps = ReadVideo(filepath) 33 | if self.merge_audio: 34 | p = subprocess.Popen(f'ffmpeg -i {filepath} -f mp3 {self.tmp_audio_path}') 35 | while True: 36 | if subprocess.Popen.poll(p) is not None: break 37 | # --不支持的数据格式 38 | else: 39 | raise RuntimeError('Unsupport file type %s...' % filepath.split('.')[-1]) 40 | outputs, pbar = [], tqdm(images) 41 | for image in pbar: 42 | pbar.set_description('Process image') 43 | outputs.append(self.iterimage(image)) 44 | if len(outputs) > 1: 45 | fps, ext = 25, 'avi' 46 | if hasattr(self, 'fps'): fps = self.fps 47 | if hasattr(self, 'ext'): ext = self.ext 48 | Images2VideoAndSave(outputs, savedir=self.savedir, savename=self.savename, fps=fps, ext=ext, logger_handle=self.logger_handle) 49 | else: 50 | ext = 'png' 51 | if hasattr(self, 'ext'): ext = self.ext 52 | SaveImage(outputs[0], savedir=self.savedir, savename=self.savename, ext=ext, logger_handle=self.logger_handle) 53 | # 如果有音频, 则把音频merge进新视频 54 | if self.merge_audio: 55 | p = subprocess.Popen(f'ffmpeg -i {os.path.join(self.savedir, self.savename+f".{ext}")} -i {self.tmp_audio_path} -strict -2 -f mp4 {os.path.join(self.savedir, self.savename+"_audio.mp4")}') 56 | while True: 57 | if subprocess.Popen.poll(p) is not None: break 58 | os.remove(self.tmp_audio_path) 59 | self.logger_handle.info(f'Video with merged audio is saved into {os.path.join(self.savedir, self.savename+"_audio.mp4")}') 60 | '''迭代图片''' 61 | def iterimage(self, image): 62 | raise NotImplementedError('not to be implemented') -------------------------------------------------------------------------------- /pydrawing/modules/beautifiers/beziercurve/__init__.py: -------------------------------------------------------------------------------- 1 | '''initialize''' 2 | from .beziercurve import BezierCurveBeautifier -------------------------------------------------------------------------------- /pydrawing/modules/beautifiers/beziercurve/beziercurve.py: -------------------------------------------------------------------------------- 1 | ''' 2 | Function: 3 | 利用贝塞尔曲线画画 4 | Author: 5 | Charles 6 | 微信公众号: 7 | Charles的皮卡丘 8 | ''' 9 | import os 10 | import re 11 | import cv2 12 | import turtle 13 | import numpy as np 14 | from bs4 import BeautifulSoup 15 | from ..base import BaseBeautifier 16 | 17 | 18 | '''利用贝塞尔曲线画画''' 19 | class BezierCurveBeautifier(BaseBeautifier): 20 | def __init__(self, num_samples=15, width=600, height=600, num_colors=32, **kwargs): 21 | super(BezierCurveBeautifier, self).__init__(**kwargs) 22 | self.num_samples = num_samples 23 | self.width = width 24 | self.height = height 25 | self.num_colors = num_colors 26 | self.rootdir = os.path.split(os.path.abspath(__file__))[0] 27 | '''迭代图片''' 28 | def iterimage(self, image): 29 | data = image.reshape((-1, 3)) 30 | data = np.float32(data) 31 | # 聚类迭代停止的模式(停止的条件, 迭代最大次数, 精度) 32 | criteria = (cv2.TERM_CRITERIA_EPS, 10, 1.0) 33 | # 数据, 分类数, 预设的分类标签, 迭代停止的模式, 重复试验kmeans算法次数, 初始类中心的选择方式 34 | compactness, labels, centers = cv2.kmeans(data, self.num_colors, None, criteria, 10, cv2.KMEANS_RANDOM_CENTERS) 35 | centers = np.uint8(centers) 36 | data_compress = centers[labels.flatten()] 37 | img_new = data_compress.reshape(image.shape) 38 | count = 0 39 | for center in centers: 40 | count += 1 41 | part = cv2.inRange(img_new, center, center) 42 | part = cv2.bitwise_not(part) 43 | cv2.imwrite('.tmp.bmp', part) 44 | os.system(f'{os.path.join(self.rootdir, "potrace.exe")} .tmp.bmp -s --flat') 45 | if count == 1: 46 | self.drawsvg('.tmp.svg', '#%02x%02x%02x' % (center[2], center[1], center[0]), True) 47 | else: 48 | self.drawsvg('.tmp.svg', '#%02x%02x%02x' % (center[2], center[1], center[0]), False) 49 | os.remove('.tmp.bmp') 50 | os.remove('.tmp.svg') 51 | turtle.done() 52 | return image 53 | '''一阶贝塞尔''' 54 | def FirstOrderBezier(self, p0, p1, t): 55 | assert (len(p0) == 2 and len(p1) == 2) or (len(p0) == 1 and len(p1) == 1) 56 | if len(p0) == 2 and len(p1) == 2: 57 | return p0[0] * (1 - t) + p1[0] * t, p0[1] * (1 - t) + p1[1] * t 58 | else: 59 | return p0 * (1 - t) + p1 * t 60 | '''二阶贝塞尔''' 61 | def SecondOrderBezier(self, p0, p1, p2): 62 | turtle.goto(p0) 63 | turtle.pendown() 64 | for t in range(0, self.num_samples+1): 65 | p = self.FirstOrderBezier(self.FirstOrderBezier(p0, p1, t/self.num_samples), self.FirstOrderBezier(p1, p2, t/self.num_samples), t/self.num_samples) 66 | turtle.goto(p) 67 | turtle.penup() 68 | '''三阶贝塞尔''' 69 | def ThirdOrderBezier(self, p0, p1, p2, p3): 70 | p0 = -self.width / 2 + p0[0], self.height / 2 - p0[1] 71 | p1 = -self.width / 2 + p1[0], self.height / 2 - p1[1] 72 | p2 = -self.width / 2 + p2[0], self.height / 2 - p2[1] 73 | p3 = -self.width / 2 + p3[0], self.height / 2 - p3[1] 74 | turtle.goto(p0) 75 | turtle.pendown() 76 | for t in range(0, self.num_samples+1): 77 | p = self.FirstOrderBezier( 78 | self.FirstOrderBezier(self.FirstOrderBezier(p0, p1, t/self.num_samples), self.FirstOrderBezier(p1, p2, t/self.num_samples), t/self.num_samples), 79 | self.FirstOrderBezier(self.FirstOrderBezier(p1, p2, t/self.num_samples), self.FirstOrderBezier(p2, p3, t/self.num_samples), t/self.num_samples), 80 | t/self.num_samples 81 | ) 82 | turtle.goto(p) 83 | turtle.penup() 84 | '''画图(SVG)''' 85 | def drawsvg(self, filename, color, is_first=True, speed=1000): 86 | svgfile = open(filename, 'r') 87 | soup = BeautifulSoup(svgfile.read(), 'lxml') 88 | height, width = float(soup.svg.attrs['height'][:-2]), float(soup.svg.attrs['width'][:-2]) 89 | scale = tuple(map(float, re.findall(r'scale\((.*?)\)', soup.g.attrs['transform'])[0].split(','))) 90 | scale = scale[0], -scale[1] 91 | if is_first: 92 | turtle.setup(height=height, width=width) 93 | turtle.setworldcoordinates(-width/2, 300, width-width/2, -height+300) 94 | turtle.tracer(100) 95 | turtle.pensize(1) 96 | turtle.speed(speed) 97 | turtle.penup() 98 | turtle.color(color) 99 | for path in soup.find_all('path'): 100 | attrs = path.attrs['d'].replace('\n', ' ') 101 | attrs = attrs.split(' ') 102 | attrs_yield = self.yieldattrs(attrs) 103 | endl = '' 104 | for attr in attrs_yield: 105 | if attr == 'M': 106 | turtle.end_fill() 107 | x, y = attrs_yield.__next__() * scale[0], attrs_yield.__next__() * scale[1] 108 | turtle.penup() 109 | turtle.goto(-self.width/2+x, self.height/2-y) 110 | turtle.pendown() 111 | turtle.begin_fill() 112 | elif attr == 'm': 113 | turtle.end_fill() 114 | dx, dy = attrs_yield.__next__() * scale[0], attrs_yield.__next__() * scale[1] 115 | turtle.penup() 116 | turtle.goto(turtle.xcor()+dx, turtle.ycor()-dy) 117 | turtle.pendown() 118 | turtle.begin_fill() 119 | elif attr == 'C': 120 | p1 = attrs_yield.__next__() * scale[0], attrs_yield.__next__() * scale[1] 121 | p2 = attrs_yield.__next__() * scale[0], attrs_yield.__next__() * scale[1] 122 | p3 = attrs_yield.__next__() * scale[0], attrs_yield.__next__() * scale[1] 123 | turtle.penup() 124 | p0 = turtle.xcor() + self.width / 2, self.height / 2 - turtle.ycor() 125 | self.ThirdOrderBezier(p0, p1, p2, p3) 126 | endl = attr 127 | elif attr == 'c': 128 | turtle.penup() 129 | p0 = turtle.xcor() + self.width / 2, self.height / 2 - turtle.ycor() 130 | p1 = attrs_yield.__next__() * scale[0] + p0[0], attrs_yield.__next__() * scale[1] + p0[1] 131 | p2 = attrs_yield.__next__() * scale[0] + p0[0], attrs_yield.__next__() * scale[1] + p0[1] 132 | p3 = attrs_yield.__next__() * scale[0] + p0[0], attrs_yield.__next__() * scale[1] + p0[1] 133 | self.ThirdOrderBezier(p0, p1, p2, p3) 134 | endl = attr 135 | elif attr == 'L': 136 | x, y = attrs_yield.__next__() * scale[0], attrs_yield.__next__() * scale[1] 137 | turtle.pendown() 138 | turtle.goto(-self.width/2+x, self.height/2-y) 139 | turtle.penup() 140 | elif attr == 'l': 141 | dx, dy = attrs_yield.__next__() * scale[0], attrs_yield.__next__() * scale[1] 142 | turtle.pendown() 143 | turtle.goto(turtle.xcor()+dx, turtle.ycor()-dy) 144 | turtle.penup() 145 | endl = attr 146 | elif endl == 'C': 147 | p1 = attr * scale[0], attrs_yield.__next__() * scale[1] 148 | p2 = attrs_yield.__next__() * scale[0], attrs_yield.__next__() * scale[1] 149 | p3 = attrs_yield.__next__() * scale[0], attrs_yield.__next__() * scale[1] 150 | turtle.penup() 151 | p0 = turtle.xcor() + self.width / 2, self.height / 2 - turtle.ycor() 152 | self.ThirdOrderBezier(p0, p1, p2, p3) 153 | elif endl == 'c': 154 | turtle.penup() 155 | p0 = turtle.xcor() + self.width / 2, self.height / 2 - turtle.ycor() 156 | p1 = attr * scale[0] + p0[0], attrs_yield.__next__() * scale[1] + p0[1] 157 | p2 = attrs_yield.__next__() * scale[0] + p0[0], attrs_yield.__next__() * scale[1] + p0[1] 158 | p3 = attrs_yield.__next__() * scale[0] + p0[0], attrs_yield.__next__() * scale[1] + p0[1] 159 | self.ThirdOrderBezier(p0, p1, p2, p3) 160 | elif endl == 'L': 161 | x, y = attr * scale[0], attrs_yield.__next__() * scale[1] 162 | turtle.pendown() 163 | turtle.goto(-self.width/2+x, self.height/2-y) 164 | turtle.penup() 165 | elif endl == 'l': 166 | dx, dy = attr * scale[0], attrs_yield.__next__() * scale[1] 167 | turtle.pendown() 168 | turtle.goto(turtle.xcor()+dx, turtle.ycor()-dy) 169 | turtle.penup() 170 | turtle.penup() 171 | turtle.hideturtle() 172 | turtle.update() 173 | svgfile.close() 174 | '''attrs生成器''' 175 | @staticmethod 176 | def yieldattrs(attrs): 177 | for attr in attrs: 178 | if attr.isdigit(): 179 | yield float(attr) 180 | elif attr[0].isalpha(): 181 | yield attr[0] 182 | yield float(attr[1:]) 183 | elif attr[-1].isalpha(): 184 | yield float(attr[0: -1]) 185 | elif attr[0] == '-': 186 | yield float(attr) -------------------------------------------------------------------------------- /pydrawing/modules/beautifiers/beziercurve/potrace.exe: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/CharlesPikachu/pydrawing/e980ad9bf4cece42ff40ed2bc7bed7a155eabb8e/pydrawing/modules/beautifiers/beziercurve/potrace.exe -------------------------------------------------------------------------------- /pydrawing/modules/beautifiers/cartoongan/__init__.py: -------------------------------------------------------------------------------- 1 | '''initialize''' 2 | from .cartoongan import CartoonGanBeautifier -------------------------------------------------------------------------------- /pydrawing/modules/beautifiers/cartoongan/cartoongan.py: -------------------------------------------------------------------------------- 1 | ''' 2 | Function: 3 | 复现论文"CartoonGAN: Generative Adversarial Networks for Photo Cartoonization" 4 | Author: 5 | Charles 6 | 微信公众号: 7 | Charles的皮卡丘 8 | ''' 9 | import cv2 10 | import numpy as np 11 | from ..base import BaseBeautifier 12 | try: 13 | import torch 14 | import torch.nn as nn 15 | import torch.nn.functional as F 16 | import torch.utils.model_zoo as model_zoo 17 | import torchvision.transforms as transforms 18 | except: 19 | print('[Warning]: Pytorch and torchvision have not be installed, "cartoongan" will be not available.') 20 | 21 | 22 | '''instance normalization''' 23 | class InstanceNormalization(nn.Module): 24 | def __init__(self, dim, eps=1e-9): 25 | super(InstanceNormalization, self).__init__() 26 | self.scale = nn.Parameter(torch.FloatTensor(dim)) 27 | self.shift = nn.Parameter(torch.FloatTensor(dim)) 28 | self.eps = eps 29 | '''call''' 30 | def __call__(self, x): 31 | n = x.size(2) * x.size(3) 32 | t = x.view(x.size(0), x.size(1), n) 33 | mean = torch.mean(t, 2).unsqueeze(2).unsqueeze(3).expand_as(x) 34 | var = torch.var(t, 2).unsqueeze(2).unsqueeze(3).expand_as(x) * ((n - 1) / float(n)) 35 | scale_broadcast = self.scale.unsqueeze(1).unsqueeze(1).unsqueeze(0) 36 | scale_broadcast = scale_broadcast.expand_as(x) 37 | shift_broadcast = self.shift.unsqueeze(1).unsqueeze(1).unsqueeze(0) 38 | shift_broadcast = shift_broadcast.expand_as(x) 39 | out = (x - mean) / torch.sqrt(var + self.eps) 40 | out = out * scale_broadcast + shift_broadcast 41 | return out 42 | 43 | 44 | '''网络结构, 模型修改自: https://github.com/Yijunmaverick/CartoonGAN-Test-Pytorch-Torch''' 45 | class Transformer(nn.Module): 46 | def __init__(self): 47 | super(Transformer, self).__init__() 48 | self.refpad01_1 = nn.ReflectionPad2d(3) 49 | self.conv01_1 = nn.Conv2d(3, 64, 7) 50 | self.in01_1 = InstanceNormalization(64) 51 | self.conv02_1 = nn.Conv2d(64, 128, 3, 2, 1) 52 | self.conv02_2 = nn.Conv2d(128, 128, 3, 1, 1) 53 | self.in02_1 = InstanceNormalization(128) 54 | self.conv03_1 = nn.Conv2d(128, 256, 3, 2, 1) 55 | self.conv03_2 = nn.Conv2d(256, 256, 3, 1, 1) 56 | self.in03_1 = InstanceNormalization(256) 57 | self.refpad04_1 = nn.ReflectionPad2d(1) 58 | self.conv04_1 = nn.Conv2d(256, 256, 3) 59 | self.in04_1 = InstanceNormalization(256) 60 | self.refpad04_2 = nn.ReflectionPad2d(1) 61 | self.conv04_2 = nn.Conv2d(256, 256, 3) 62 | self.in04_2 = InstanceNormalization(256) 63 | self.refpad05_1 = nn.ReflectionPad2d(1) 64 | self.conv05_1 = nn.Conv2d(256, 256, 3) 65 | self.in05_1 = InstanceNormalization(256) 66 | self.refpad05_2 = nn.ReflectionPad2d(1) 67 | self.conv05_2 = nn.Conv2d(256, 256, 3) 68 | self.in05_2 = InstanceNormalization(256) 69 | self.refpad06_1 = nn.ReflectionPad2d(1) 70 | self.conv06_1 = nn.Conv2d(256, 256, 3) 71 | self.in06_1 = InstanceNormalization(256) 72 | self.refpad06_2 = nn.ReflectionPad2d(1) 73 | self.conv06_2 = nn.Conv2d(256, 256, 3) 74 | self.in06_2 = InstanceNormalization(256) 75 | self.refpad07_1 = nn.ReflectionPad2d(1) 76 | self.conv07_1 = nn.Conv2d(256, 256, 3) 77 | self.in07_1 = InstanceNormalization(256) 78 | self.refpad07_2 = nn.ReflectionPad2d(1) 79 | self.conv07_2 = nn.Conv2d(256, 256, 3) 80 | self.in07_2 = InstanceNormalization(256) 81 | self.refpad08_1 = nn.ReflectionPad2d(1) 82 | self.conv08_1 = nn.Conv2d(256, 256, 3) 83 | self.in08_1 = InstanceNormalization(256) 84 | self.refpad08_2 = nn.ReflectionPad2d(1) 85 | self.conv08_2 = nn.Conv2d(256, 256, 3) 86 | self.in08_2 = InstanceNormalization(256) 87 | self.refpad09_1 = nn.ReflectionPad2d(1) 88 | self.conv09_1 = nn.Conv2d(256, 256, 3) 89 | self.in09_1 = InstanceNormalization(256) 90 | self.refpad09_2 = nn.ReflectionPad2d(1) 91 | self.conv09_2 = nn.Conv2d(256, 256, 3) 92 | self.in09_2 = InstanceNormalization(256) 93 | self.refpad10_1 = nn.ReflectionPad2d(1) 94 | self.conv10_1 = nn.Conv2d(256, 256, 3) 95 | self.in10_1 = InstanceNormalization(256) 96 | self.refpad10_2 = nn.ReflectionPad2d(1) 97 | self.conv10_2 = nn.Conv2d(256, 256, 3) 98 | self.in10_2 = InstanceNormalization(256) 99 | self.refpad11_1 = nn.ReflectionPad2d(1) 100 | self.conv11_1 = nn.Conv2d(256, 256, 3) 101 | self.in11_1 = InstanceNormalization(256) 102 | self.refpad11_2 = nn.ReflectionPad2d(1) 103 | self.conv11_2 = nn.Conv2d(256, 256, 3) 104 | self.in11_2 = InstanceNormalization(256) 105 | self.deconv01_1 = nn.ConvTranspose2d(256, 128, 3, 2, 1, 1) 106 | self.deconv01_2 = nn.Conv2d(128, 128, 3, 1, 1) 107 | self.in12_1 = InstanceNormalization(128) 108 | self.deconv02_1 = nn.ConvTranspose2d(128, 64, 3, 2, 1, 1) 109 | self.deconv02_2 = nn.Conv2d(64, 64, 3, 1, 1) 110 | self.in13_1 = InstanceNormalization(64) 111 | self.refpad12_1 = nn.ReflectionPad2d(3) 112 | self.deconv03_1 = nn.Conv2d(64, 3, 7) 113 | '''forward''' 114 | def forward(self, x): 115 | y = F.relu(self.in01_1(self.conv01_1(self.refpad01_1(x)))) 116 | y = F.relu(self.in02_1(self.conv02_2(self.conv02_1(y)))) 117 | t04 = F.relu(self.in03_1(self.conv03_2(self.conv03_1(y)))) 118 | y = F.relu(self.in04_1(self.conv04_1(self.refpad04_1(t04)))) 119 | t05 = self.in04_2(self.conv04_2(self.refpad04_2(y))) + t04 120 | y = F.relu(self.in05_1(self.conv05_1(self.refpad05_1(t05)))) 121 | t06 = self.in05_2(self.conv05_2(self.refpad05_2(y))) + t05 122 | y = F.relu(self.in06_1(self.conv06_1(self.refpad06_1(t06)))) 123 | t07 = self.in06_2(self.conv06_2(self.refpad06_2(y))) + t06 124 | y = F.relu(self.in07_1(self.conv07_1(self.refpad07_1(t07)))) 125 | t08 = self.in07_2(self.conv07_2(self.refpad07_2(y))) + t07 126 | y = F.relu(self.in08_1(self.conv08_1(self.refpad08_1(t08)))) 127 | t09 = self.in08_2(self.conv08_2(self.refpad08_2(y))) + t08 128 | y = F.relu(self.in09_1(self.conv09_1(self.refpad09_1(t09)))) 129 | t10 = self.in09_2(self.conv09_2(self.refpad09_2(y))) + t09 130 | y = F.relu(self.in10_1(self.conv10_1(self.refpad10_1(t10)))) 131 | t11 = self.in10_2(self.conv10_2(self.refpad10_2(y))) + t10 132 | y = F.relu(self.in11_1(self.conv11_1(self.refpad11_1(t11)))) 133 | y = self.in11_2(self.conv11_2(self.refpad11_2(y))) + t11 134 | y = F.relu(self.in12_1(self.deconv01_2(self.deconv01_1(y)))) 135 | y = F.relu(self.in13_1(self.deconv02_2(self.deconv02_1(y)))) 136 | y = F.tanh(self.deconv03_1(self.refpad12_1(y))) 137 | return y 138 | 139 | 140 | '''复现论文"CartoonGAN: Generative Adversarial Networks for Photo Cartoonization"''' 141 | class CartoonGanBeautifier(BaseBeautifier): 142 | def __init__(self, style='Hosoda', use_cuda=True, **kwargs): 143 | super(CartoonGanBeautifier, self).__init__(**kwargs) 144 | self.model_urls = { 145 | 'Hayao': 'http://vllab1.ucmerced.edu/~yli62/CartoonGAN/pytorch_pth/Hayao_net_G_float.pth', 146 | 'Hosoda': 'http://vllab1.ucmerced.edu/~yli62/CartoonGAN/pytorch_pth/Hosoda_net_G_float.pth', 147 | 'Paprika': 'http://vllab1.ucmerced.edu/~yli62/CartoonGAN/pytorch_pth/Paprika_net_G_float.pth', 148 | 'Shinkai': 'http://vllab1.ucmerced.edu/~yli62/CartoonGAN/pytorch_pth/Shinkai_net_G_float.pth', 149 | } 150 | assert style in self.model_urls 151 | self.style = style 152 | self.use_cuda = use_cuda 153 | self.transformer = Transformer() 154 | self.transformer.load_state_dict(model_zoo.load_url(self.model_urls[style], map_location='cpu')) 155 | self.transformer.eval() 156 | if torch.cuda.is_available() and self.use_cuda: self.transformer = self.transformer.cuda() 157 | '''迭代图片''' 158 | def iterimage(self, image): 159 | input_image = transforms.ToTensor()(image).unsqueeze(0) 160 | input_image = -1 + 2 * input_image 161 | if torch.cuda.is_available() and self.use_cuda: 162 | input_image = input_image.cuda() 163 | with torch.no_grad(): 164 | output_image = self.transformer(input_image)[0] 165 | output_image = output_image.data.cpu().float() * 0.5 + 0.5 166 | output_image = (output_image.numpy() * 255).astype(np.uint8) 167 | output_image = np.transpose(output_image, (1, 2, 0)) 168 | output_image = cv2.resize(output_image, (image.shape[1], image.shape[0])) 169 | return output_image -------------------------------------------------------------------------------- /pydrawing/modules/beautifiers/cartoonise/__init__.py: -------------------------------------------------------------------------------- 1 | '''initialize''' 2 | from .cartoonise import CartooniseBeautifier -------------------------------------------------------------------------------- /pydrawing/modules/beautifiers/cartoonise/cartoonise.py: -------------------------------------------------------------------------------- 1 | ''' 2 | Function: 3 | 图像卡通化 4 | Author: 5 | Charles 6 | 微信公众号: 7 | Charles的皮卡丘 8 | ''' 9 | import cv2 10 | import numpy as np 11 | from ..base import BaseBeautifier 12 | 13 | 14 | '''图像卡通化''' 15 | class CartooniseBeautifier(BaseBeautifier): 16 | def __init__(self, mode='rgb', **kwargs): 17 | super(CartooniseBeautifier, self).__init__(**kwargs) 18 | assert mode in ['rgb', 'hsv'] 19 | self.mode = mode 20 | '''迭代图片''' 21 | def iterimage(self, image): 22 | if self.mode == 'rgb': 23 | return self.processinrgb(image) 24 | elif self.mode == 'hsv': 25 | return self.processinhsv(image) 26 | '''在RGB空间操作''' 27 | def processinrgb(self, image): 28 | # Step1: 利用双边滤波器对原图像进行保边去噪处理 29 | # --下采样 30 | image_bilateral = image 31 | for _ in range(2): 32 | image_bilateral = cv2.pyrDown(image_bilateral) 33 | # --进行多次的双边滤波 34 | for _ in range(7): 35 | image_bilateral = cv2.bilateralFilter(image_bilateral, d=9, sigmaColor=9, sigmaSpace=7) 36 | # --上采样 37 | for _ in range(2): 38 | image_bilateral = cv2.pyrUp(image_bilateral) 39 | # Step2: 将步骤一中获得的图像灰度化后，使用中值滤波器去噪 40 | image_gray = cv2.cvtColor(image_bilateral, cv2.COLOR_RGB2GRAY) 41 | image_median = cv2.medianBlur(image_gray, 7) 42 | # Step3: 对步骤二中获得的图像使用自适应阈值从而获得原图像的轮廓 43 | image_edge = cv2.adaptiveThreshold(image_median, 255, cv2.ADAPTIVE_THRESH_MEAN_C, cv2.THRESH_BINARY, blockSize=9, C=2) 44 | image_edge = cv2.cvtColor(image_edge, cv2.COLOR_GRAY2RGB) 45 | # Step4: 将步骤一中获得的图像与步骤三中获得的图像轮廓合并即可实现将照片变为卡通图片的效果了 46 | image_cartoon = cv2.bitwise_and(image_bilateral, image_edge) 47 | # 返回 48 | return image_cartoon 49 | '''在HSV空间操作''' 50 | def processinhsv(self, image): 51 | # Step1: 图像BGR空间转HSV空间, 在HSV空间进行直方图均衡化, 中值滤波和形态学变换 52 | image_hsv = cv2.cvtColor(image, cv2.COLOR_BGR2HSV) 53 | h, s, v = cv2.split(image_hsv) 54 | # --直方图均衡化 55 | v = cv2.equalizeHist(v) 56 | image_hsv = cv2.merge((h, s, v)) 57 | # --中值滤波 58 | image_hsv = cv2.medianBlur(image_hsv, 7) 59 | # --形态学变换-开/闭运算 60 | kernel = np.ones((5, 5), np.uint8) 61 | image_hsv = cv2.morphologyEx(image_hsv, cv2.MORPH_CLOSE, kernel, iterations=2) 62 | # --中值滤波 63 | image_hsv = cv2.medianBlur(image_hsv, 7) 64 | # Step2: 对步骤一中获得的图像使用自适应阈值从而获得原图像的轮廓 65 | image_mask = cv2.cvtColor(image_hsv, cv2.COLOR_HSV2BGR) 66 | image_mask = cv2.cvtColor(image_mask, cv2.COLOR_RGB2GRAY) 67 | image_mask = cv2.medianBlur(image_mask, 7) 68 | image_edge = cv2.adaptiveThreshold(image_mask, 255, cv2.ADAPTIVE_THRESH_MEAN_C, cv2.THRESH_BINARY, blockSize=9, C=2) 69 | image_edge = cv2.cvtColor(image_edge, cv2.COLOR_GRAY2RGB) 70 | # Step3: 将步骤二中获得的图像轮廓与原图像合并即可实现将照片变为卡通图片的效果了 71 | image_cartoon = cv2.bitwise_and(image, image_edge) 72 | # 返回 73 | return image_cartoon -------------------------------------------------------------------------------- /pydrawing/modules/beautifiers/cartoonizeface/__init__.py: -------------------------------------------------------------------------------- 1 | '''initialize''' 2 | from .cartoonizeface import CartoonizeFaceBeautifier -------------------------------------------------------------------------------- /pydrawing/modules/beautifiers/cartoonizeface/cartoonizeface.py: -------------------------------------------------------------------------------- 1 | ''' 2 | Function: 3 | 人脸卡通化 4 | Author: 5 | Charles 6 | 微信公众号: 7 | Charles的皮卡丘 8 | ''' 9 | import cv2 10 | import numpy as np 11 | from PIL import Image 12 | from ..base import BaseBeautifier 13 | try: 14 | import torch 15 | import torch.nn as nn 16 | import torch.nn.functional as F 17 | import torch.utils.model_zoo as model_zoo 18 | from torch.nn.parameter import Parameter 19 | except: 20 | print('[Warning]: Pytorch and torchvision have not be installed, "cartoonizeface" will be not available.') 21 | 22 | 23 | '''ConvBlock''' 24 | class ConvBlock(nn.Module): 25 | def __init__(self, dim_in, dim_out): 26 | super(ConvBlock, self).__init__() 27 | self.dim_out = dim_out 28 | self.ConvBlock1 = nn.Sequential( 29 | nn.InstanceNorm2d(dim_in), 30 | nn.ReLU(True), 31 | nn.ReflectionPad2d(1), 32 | nn.Conv2d(dim_in, dim_out//2, kernel_size=3, stride=1, bias=False) 33 | ) 34 | self.ConvBlock2 = nn.Sequential( 35 | nn.InstanceNorm2d(dim_out//2), 36 | nn.ReLU(True), 37 | nn.ReflectionPad2d(1), 38 | nn.Conv2d(dim_out//2, dim_out//4, kernel_size=3, stride=1, bias=False) 39 | ) 40 | self.ConvBlock3 = nn.Sequential( 41 | nn.InstanceNorm2d(dim_out//4), 42 | nn.ReLU(True), 43 | nn.ReflectionPad2d(1), 44 | nn.Conv2d(dim_out//4, dim_out//4, kernel_size=3, stride=1, bias=False) 45 | ) 46 | self.ConvBlock4 = nn.Sequential( 47 | nn.InstanceNorm2d(dim_in), 48 | nn.ReLU(True), 49 | nn.Conv2d(dim_in, dim_out, kernel_size=1, stride=1, bias=False) 50 | ) 51 | '''forward''' 52 | def forward(self, x): 53 | residual = x 54 | x1 = self.ConvBlock1(x) 55 | x2 = self.ConvBlock2(x1) 56 | x3 = self.ConvBlock3(x2) 57 | out = torch.cat((x1, x2, x3), 1) 58 | if residual.size(1) != self.dim_out: residual = self.ConvBlock4(residual) 59 | return residual + out 60 | 61 | 62 | '''HourGlassBlock''' 63 | class HourGlassBlock(nn.Module): 64 | def __init__(self, dim_in, dim_out): 65 | super(HourGlassBlock, self).__init__() 66 | self.ConvBlock1_1 = ConvBlock(dim_in, dim_out) 67 | self.ConvBlock1_2 = ConvBlock(dim_out, dim_out) 68 | self.ConvBlock2_1 = ConvBlock(dim_out, dim_out) 69 | self.ConvBlock2_2 = ConvBlock(dim_out, dim_out) 70 | self.ConvBlock3_1 = ConvBlock(dim_out, dim_out) 71 | self.ConvBlock3_2 = ConvBlock(dim_out, dim_out) 72 | self.ConvBlock4_1 = ConvBlock(dim_out, dim_out) 73 | self.ConvBlock4_2 = ConvBlock(dim_out, dim_out) 74 | self.ConvBlock5 = ConvBlock(dim_out, dim_out) 75 | self.ConvBlock6 = ConvBlock(dim_out, dim_out) 76 | self.ConvBlock7 = ConvBlock(dim_out, dim_out) 77 | self.ConvBlock8 = ConvBlock(dim_out, dim_out) 78 | self.ConvBlock9 = ConvBlock(dim_out, dim_out) 79 | '''forward''' 80 | def forward(self, x): 81 | skip1 = self.ConvBlock1_1(x) 82 | down1 = F.avg_pool2d(x, 2) 83 | down1 = self.ConvBlock1_2(down1) 84 | skip2 = self.ConvBlock2_1(down1) 85 | down2 = F.avg_pool2d(down1, 2) 86 | down2 = self.ConvBlock2_2(down2) 87 | skip3 = self.ConvBlock3_1(down2) 88 | down3 = F.avg_pool2d(down2, 2) 89 | down3 = self.ConvBlock3_2(down3) 90 | skip4 = self.ConvBlock4_1(down3) 91 | down4 = F.avg_pool2d(down3, 2) 92 | down4 = self.ConvBlock4_2(down4) 93 | center = self.ConvBlock5(down4) 94 | up4 = self.ConvBlock6(center) 95 | up4 = F.upsample(up4, scale_factor=2) 96 | up4 = skip4 + up4 97 | up3 = self.ConvBlock7(up4) 98 | up3 = F.upsample(up3, scale_factor=2) 99 | up3 = skip3 + up3 100 | up2 = self.ConvBlock8(up3) 101 | up2 = F.upsample(up2, scale_factor=2) 102 | up2 = skip2 + up2 103 | up1 = self.ConvBlock9(up2) 104 | up1 = F.upsample(up1, scale_factor=2) 105 | up1 = skip1 + up1 106 | return up1 107 | 108 | 109 | '''HourGlass''' 110 | class HourGlass(nn.Module): 111 | def __init__(self, dim_in, dim_out, use_res=True): 112 | super(HourGlass, self).__init__() 113 | self.use_res = use_res 114 | self.HG = nn.Sequential( 115 | HourGlassBlock(dim_in, dim_out), 116 | ConvBlock(dim_out, dim_out), 117 | nn.Conv2d(dim_out, dim_out, kernel_size=1, stride=1, bias=False), 118 | nn.InstanceNorm2d(dim_out), 119 | nn.ReLU(True) 120 | ) 121 | self.Conv1 = nn.Conv2d(dim_out, 3, kernel_size=1, stride=1) 122 | if self.use_res: 123 | self.Conv2 = nn.Conv2d(dim_out, dim_out, kernel_size=1, stride=1) 124 | self.Conv3 = nn.Conv2d(3, dim_out, kernel_size=1, stride=1) 125 | '''forward''' 126 | def forward(self, x): 127 | ll = self.HG(x) 128 | tmp_out = self.Conv1(ll) 129 | if self.use_res: 130 | ll = self.Conv2(ll) 131 | tmp_out_ = self.Conv3(tmp_out) 132 | return x + ll + tmp_out_ 133 | else: 134 | return tmp_out 135 | 136 | 137 | '''ResnetBlock''' 138 | class ResnetBlock(nn.Module): 139 | def __init__(self, dim, use_bias=False): 140 | super(ResnetBlock, self).__init__() 141 | conv_block = [] 142 | conv_block += [nn.ReflectionPad2d(1), nn.Conv2d(dim, dim, kernel_size=3, stride=1, padding=0, bias=use_bias), nn.InstanceNorm2d(dim), nn.ReLU(True)] 143 | conv_block += [nn.ReflectionPad2d(1), nn.Conv2d(dim, dim, kernel_size=3, stride=1, padding=0, bias=use_bias), nn.InstanceNorm2d(dim)] 144 | self.conv_block = nn.Sequential(*conv_block) 145 | '''forward''' 146 | def forward(self, x): 147 | out = x + self.conv_block(x) 148 | return out 149 | 150 | 151 | '''adaLIN''' 152 | class adaLIN(nn.Module): 153 | def __init__(self, num_features, eps=1e-5): 154 | super(adaLIN, self).__init__() 155 | self.eps = eps 156 | self.rho = Parameter(torch.Tensor(1, num_features, 1, 1)) 157 | self.rho.data.fill_(0.9) 158 | '''forward''' 159 | def forward(self, input, gamma, beta): 160 | in_mean, in_var = torch.mean(input, dim=[2, 3], keepdim=True), torch.var(input, dim=[2, 3], keepdim=True) 161 | out_in = (input - in_mean) / torch.sqrt(in_var + self.eps) 162 | ln_mean, ln_var = torch.mean(input, dim=[1, 2, 3], keepdim=True), torch.var(input, dim=[1, 2, 3], keepdim=True) 163 | out_ln = (input - ln_mean) / torch.sqrt(ln_var + self.eps) 164 | out = self.rho.expand(input.shape[0], -1, -1, -1) * out_in + (1-self.rho.expand(input.shape[0], -1, -1, -1)) * out_ln 165 | out = out * gamma.unsqueeze(2).unsqueeze(3) + beta.unsqueeze(2).unsqueeze(3) 166 | return out 167 | 168 | 169 | '''SoftAdaLIN''' 170 | class SoftAdaLIN(nn.Module): 171 | def __init__(self, num_features, eps=1e-5): 172 | super(SoftAdaLIN, self).__init__() 173 | self.norm = adaLIN(num_features, eps) 174 | self.w_gamma = Parameter(torch.zeros(1, num_features)) 175 | self.w_beta = Parameter(torch.zeros(1, num_features)) 176 | self.c_gamma = nn.Sequential(nn.Linear(num_features, num_features), nn.ReLU(True), nn.Linear(num_features, num_features)) 177 | self.c_beta = nn.Sequential(nn.Linear(num_features, num_features), nn.ReLU(True), nn.Linear(num_features, num_features)) 178 | self.s_gamma = nn.Linear(num_features, num_features) 179 | self.s_beta = nn.Linear(num_features, num_features) 180 | '''forward''' 181 | def forward(self, x, content_features, style_features): 182 | content_gamma, content_beta = self.c_gamma(content_features), self.c_beta(content_features) 183 | style_gamma, style_beta = self.s_gamma(style_features), self.s_beta(style_features) 184 | w_gamma, w_beta = self.w_gamma.expand(x.shape[0], -1), self.w_beta.expand(x.shape[0], -1) 185 | soft_gamma = (1. - w_gamma) * style_gamma + w_gamma * content_gamma 186 | soft_beta = (1. - w_beta) * style_beta + w_beta * content_beta 187 | out = self.norm(x, soft_gamma, soft_beta) 188 | return out 189 | 190 | 191 | '''ResnetSoftAdaLINBlock''' 192 | class ResnetSoftAdaLINBlock(nn.Module): 193 | def __init__(self, dim, use_bias=False): 194 | super(ResnetSoftAdaLINBlock, self).__init__() 195 | self.pad1 = nn.ReflectionPad2d(1) 196 | self.conv1 = nn.Conv2d(dim, dim, kernel_size=3, stride=1, padding=0, bias=use_bias) 197 | self.norm1 = SoftAdaLIN(dim) 198 | self.relu1 = nn.ReLU(True) 199 | self.pad2 = nn.ReflectionPad2d(1) 200 | self.conv2 = nn.Conv2d(dim, dim, kernel_size=3, stride=1, padding=0, bias=use_bias) 201 | self.norm2 = SoftAdaLIN(dim) 202 | '''forward''' 203 | def forward(self, x, content_features, style_features): 204 | out = self.pad1(x) 205 | out = self.conv1(out) 206 | out = self.norm1(out, content_features, style_features) 207 | out = self.relu1(out) 208 | out = self.pad2(out) 209 | out = self.conv2(out) 210 | out = self.norm2(out, content_features, style_features) 211 | return out + x 212 | 213 | 214 | '''LIN''' 215 | class LIN(nn.Module): 216 | def __init__(self, num_features, eps=1e-5): 217 | super(LIN, self).__init__() 218 | self.eps = eps 219 | self.rho = Parameter(torch.Tensor(1, num_features, 1, 1)) 220 | self.gamma = Parameter(torch.Tensor(1, num_features, 1, 1)) 221 | self.beta = Parameter(torch.Tensor(1, num_features, 1, 1)) 222 | self.rho.data.fill_(0.0) 223 | self.gamma.data.fill_(1.0) 224 | self.beta.data.fill_(0.0) 225 | '''forward''' 226 | def forward(self, input): 227 | in_mean, in_var = torch.mean(input, dim=[2, 3], keepdim=True), torch.var(input, dim=[2, 3], keepdim=True) 228 | out_in = (input - in_mean) / torch.sqrt(in_var + self.eps) 229 | ln_mean, ln_var = torch.mean(input, dim=[1, 2, 3], keepdim=True), torch.var(input, dim=[1, 2, 3], keepdim=True) 230 | out_ln = (input - ln_mean) / torch.sqrt(ln_var + self.eps) 231 | out = self.rho.expand(input.shape[0], -1, -1, -1) * out_in + (1-self.rho.expand(input.shape[0], -1, -1, -1)) * out_ln 232 | out = out * self.gamma.expand(input.shape[0], -1, -1, -1) + self.beta.expand(input.shape[0], -1, -1, -1) 233 | return out 234 | 235 | 236 | '''ResnetGenerator, 模型修改自: https://github.com/minivision-ai/photo2cartoon''' 237 | class ResnetGenerator(nn.Module): 238 | def __init__(self, ngf=64, img_size=256, light=False): 239 | super(ResnetGenerator, self).__init__() 240 | self.light = light 241 | self.ConvBlock1 = nn.Sequential( 242 | nn.ReflectionPad2d(3), 243 | nn.Conv2d(3, ngf, kernel_size=7, stride=1, padding=0, bias=False), 244 | nn.InstanceNorm2d(ngf), 245 | nn.ReLU(True) 246 | ) 247 | self.HourGlass1 = HourGlass(ngf, ngf) 248 | self.HourGlass2 = HourGlass(ngf, ngf) 249 | # Down-Sampling 250 | self.DownBlock1 = nn.Sequential( 251 | nn.ReflectionPad2d(1), 252 | nn.Conv2d(ngf, ngf*2, kernel_size=3, stride=2, padding=0, bias=False), 253 | nn.InstanceNorm2d(ngf * 2), 254 | nn.ReLU(True) 255 | ) 256 | self.DownBlock2 = nn.Sequential( 257 | nn.ReflectionPad2d(1), 258 | nn.Conv2d(ngf*2, ngf*4, kernel_size=3, stride=2, padding=0, bias=False), 259 | nn.InstanceNorm2d(ngf*4), 260 | nn.ReLU(True) 261 | ) 262 | # Encoder Bottleneck 263 | self.EncodeBlock1 = ResnetBlock(ngf*4) 264 | self.EncodeBlock2 = ResnetBlock(ngf*4) 265 | self.EncodeBlock3 = ResnetBlock(ngf*4) 266 | self.EncodeBlock4 = ResnetBlock(ngf*4) 267 | # Class Activation Map 268 | self.gap_fc = nn.Linear(ngf*4, 1) 269 | self.gmp_fc = nn.Linear(ngf*4, 1) 270 | self.conv1x1 = nn.Conv2d(ngf*8, ngf*4, kernel_size=1, stride=1) 271 | self.relu = nn.ReLU(True) 272 | # Gamma, Beta block 273 | if self.light: 274 | self.FC = nn.Sequential(nn.Linear(ngf*4, ngf*4), nn.ReLU(True), nn.Linear(ngf*4, ngf*4), nn.ReLU(True)) 275 | else: 276 | self.FC = nn.Sequential(nn.Linear(img_size//4*img_size//4*ngf*4, ngf*4), nn.ReLU(True), nn.Linear(ngf*4, ngf*4), nn.ReLU(True)) 277 | # Decoder Bottleneck 278 | self.DecodeBlock1 = ResnetSoftAdaLINBlock(ngf*4) 279 | self.DecodeBlock2 = ResnetSoftAdaLINBlock(ngf*4) 280 | self.DecodeBlock3 = ResnetSoftAdaLINBlock(ngf*4) 281 | self.DecodeBlock4 = ResnetSoftAdaLINBlock(ngf*4) 282 | # Up-Sampling 283 | self.UpBlock1 = nn.Sequential( 284 | nn.Upsample(scale_factor=2), 285 | nn.ReflectionPad2d(1), 286 | nn.Conv2d(ngf*4, ngf*2, kernel_size=3, stride=1, padding=0, bias=False), 287 | LIN(ngf*2), 288 | nn.ReLU(True) 289 | ) 290 | self.UpBlock2 = nn.Sequential( 291 | nn.Upsample(scale_factor=2), 292 | nn.ReflectionPad2d(1), 293 | nn.Conv2d(ngf*2, ngf, kernel_size=3, stride=1, padding=0, bias=False), 294 | LIN(ngf), 295 | nn.ReLU(True) 296 | ) 297 | self.HourGlass3 = HourGlass(ngf, ngf) 298 | self.HourGlass4 = HourGlass(ngf, ngf, False) 299 | self.ConvBlock2 = nn.Sequential( 300 | nn.ReflectionPad2d(3), 301 | nn.Conv2d(3, 3, kernel_size=7, stride=1, padding=0, bias=False), 302 | nn.Tanh() 303 | ) 304 | '''forward''' 305 | def forward(self, x): 306 | x = self.ConvBlock1(x) 307 | x = self.HourGlass1(x) 308 | x = self.HourGlass2(x) 309 | x = self.DownBlock1(x) 310 | x = self.DownBlock2(x) 311 | x = self.EncodeBlock1(x) 312 | content_features1 = F.adaptive_avg_pool2d(x, 1).view(x.shape[0], -1) 313 | x = self.EncodeBlock2(x) 314 | content_features2 = F.adaptive_avg_pool2d(x, 1).view(x.shape[0], -1) 315 | x = self.EncodeBlock3(x) 316 | content_features3 = F.adaptive_avg_pool2d(x, 1).view(x.shape[0], -1) 317 | x = self.EncodeBlock4(x) 318 | content_features4 = F.adaptive_avg_pool2d(x, 1).view(x.shape[0], -1) 319 | gap = F.adaptive_avg_pool2d(x, 1) 320 | gap_logit = self.gap_fc(gap.view(x.shape[0], -1)) 321 | gap_weight = list(self.gap_fc.parameters())[0] 322 | gap = x * gap_weight.unsqueeze(2).unsqueeze(3) 323 | gmp = F.adaptive_max_pool2d(x, 1) 324 | gmp_logit = self.gmp_fc(gmp.view(x.shape[0], -1)) 325 | gmp_weight = list(self.gmp_fc.parameters())[0] 326 | gmp = x * gmp_weight.unsqueeze(2).unsqueeze(3) 327 | cam_logit = torch.cat([gap_logit, gmp_logit], 1) 328 | x = torch.cat([gap, gmp], 1) 329 | x = self.relu(self.conv1x1(x)) 330 | heatmap = torch.sum(x, dim=1, keepdim=True) 331 | if self.light: 332 | x_ = F.adaptive_avg_pool2d(x, 1) 333 | style_features = self.FC(x_.view(x_.shape[0], -1)) 334 | else: 335 | style_features = self.FC(x.view(x.shape[0], -1)) 336 | x = self.DecodeBlock1(x, content_features4, style_features) 337 | x = self.DecodeBlock2(x, content_features3, style_features) 338 | x = self.DecodeBlock3(x, content_features2, style_features) 339 | x = self.DecodeBlock4(x, content_features1, style_features) 340 | x = self.UpBlock1(x) 341 | x = self.UpBlock2(x) 342 | x = self.HourGlass3(x) 343 | x = self.HourGlass4(x) 344 | out = self.ConvBlock2(x) 345 | return out, cam_logit, heatmap 346 | 347 | 348 | '''人脸卡通化''' 349 | class CartoonizeFaceBeautifier(BaseBeautifier): 350 | def __init__(self, use_cuda=False, use_face_segmentor=True, **kwargs): 351 | super(CartoonizeFaceBeautifier, self).__init__(**kwargs) 352 | from .facedetector import FaceDetector 353 | from .facesegmentor import FaceSegmentor 354 | self.model_urls = { 355 | 'transformer': 'https://github.com/CharlesPikachu/pydrawing/releases/download/checkpoints/cartoonizeface_transformer.pth', 356 | } 357 | self.use_cuda = use_cuda 358 | self.use_face_segmentor = use_face_segmentor 359 | self.face_detector = FaceDetector(use_cuda=(torch.cuda.is_available() and self.use_cuda)) 360 | self.transformer = ResnetGenerator(ngf=32, img_size=256, light=True) 361 | self.transformer.load_state_dict(model_zoo.load_url(self.model_urls['transformer'], map_location='cpu')['genA2B']) 362 | self.transformer.eval() 363 | if use_face_segmentor: 364 | self.face_segmentor = FaceSegmentor() 365 | self.face_segmentor.eval() 366 | if torch.cuda.is_available() and self.use_cuda: 367 | self.transformer = self.transformer.cuda() 368 | if use_face_segmentor: self.face_segmentor = self.face_segmentor.cuda() 369 | '''迭代图片''' 370 | def iterimage(self, image): 371 | image = cv2.cvtColor(image, cv2.COLOR_BGR2RGB) 372 | # 人脸提取 373 | face_rgb = self.face_detector(image) 374 | # 人脸分割 375 | if self.use_face_segmentor: 376 | face_rgb_for_seg = self.face_segmentor.preprocess(face_rgb) 377 | mask = self.face_segmentor(face_rgb_for_seg) 378 | mask = F.interpolate(mask, size=face_rgb.shape[:2][::-1], mode='bilinear', align_corners=False) 379 | mask = mask[0].argmax(0).cpu().numpy().astype(np.int32) 380 | mask = self.face_segmentor.getfacemask(mask) 381 | else: 382 | mask = np.ones(face_rgb.shape[:2]) * 255 383 | mask = mask[:, :, np.newaxis] 384 | face_rgba = np.dstack((face_rgb, mask)) 385 | # 人脸处理 386 | face_rgba = cv2.resize(face_rgba, (256, 256), interpolation=cv2.INTER_AREA) 387 | face = face_rgba[:, :, :3].copy() 388 | mask = face_rgba[:, :, 3][:, :, np.newaxis].copy() / 255. 389 | face = (face * mask + (1 - mask) * 255) / 127.5 - 1 390 | face = np.transpose(face[np.newaxis, :, :, :], (0, 3, 1, 2)).astype(np.float32) 391 | face = torch.from_numpy(face).type(torch.FloatTensor) 392 | if torch.cuda.is_available() and self.use_cuda: 393 | face = face.cuda() 394 | # 推理 395 | with torch.no_grad(): 396 | face_cartoon = self.transformer(face)[0][0] 397 | # 后处理 398 | face_cartoon = np.transpose(face_cartoon.cpu().numpy(), (1, 2, 0)) 399 | face_cartoon = (face_cartoon + 1) * 127.5 400 | face_cartoon = (face_cartoon * mask + 255 * (1 - mask)).astype(np.uint8) 401 | face_cartoon = cv2.cvtColor(face_cartoon, cv2.COLOR_RGB2BGR) 402 | # 返回 403 | return face_cartoon -------------------------------------------------------------------------------- /pydrawing/modules/beautifiers/cartoonizeface/facedetector/__init__.py: -------------------------------------------------------------------------------- 1 | '''initialize''' 2 | from .facedetector import FaceDetector -------------------------------------------------------------------------------- /pydrawing/modules/beautifiers/cartoonizeface/facedetector/facedetector.py: -------------------------------------------------------------------------------- 1 | ''' 2 | Function: 3 | 人脸检测 4 | Author: 5 | Charles 6 | 微信公众号: 7 | Charles的皮卡丘 8 | ''' 9 | import cv2 10 | import ssl 11 | import math 12 | import numpy as np 13 | ssl._create_default_https_context = ssl._create_unverified_context 14 | 15 | 16 | '''FaceDetector''' 17 | class FaceDetector(): 18 | def __init__(self, use_cuda, **kwargs): 19 | super(FaceDetector, self).__init__() 20 | try: 21 | import face_alignment 22 | except: 23 | raise RuntimeError('Please run "pip install face_alignment" to install "face_alignment"') 24 | device = 'cuda' if use_cuda else 'cpu' 25 | self.dlib_detector = face_alignment.FaceAlignment(face_alignment.LandmarksType._2D, device=device, face_detector='dlib') 26 | '''forward''' 27 | def __call__(self, image): 28 | # obtain landmarks 29 | preds = self.dlib_detector.get_landmarks(image) 30 | landmarks = None 31 | if preds is None: 32 | raise RuntimeError('no faces are detected') 33 | elif len(preds) == 1: 34 | landmarks = preds[0] 35 | else: 36 | areas = [] 37 | for pred in preds: 38 | landmarks_top = np.min(pred[:, 1]) 39 | landmarks_bottom = np.max(pred[:, 1]) 40 | landmarks_left = np.min(pred[:, 0]) 41 | landmarks_right = np.max(pred[:, 0]) 42 | areas.append((landmarks_bottom - landmarks_top) * (landmarks_right - landmarks_left)) 43 | max_face_index = np.argmax(areas) 44 | landmarks = preds[max_face_index] 45 | # rotate 46 | left_eye_corner = landmarks[36] 47 | right_eye_corner = landmarks[45] 48 | radian = np.arctan((left_eye_corner[1] - right_eye_corner[1]) / (left_eye_corner[0] - right_eye_corner[0])) 49 | height, width, _ = image.shape 50 | cos = math.cos(radian) 51 | sin = math.sin(radian) 52 | new_w = int(width * abs(cos) + height * abs(sin)) 53 | new_h = int(width * abs(sin) + height * abs(cos)) 54 | Tx = new_w // 2 - width // 2 55 | Ty = new_h // 2 - height // 2 56 | M = np.array([[cos, sin, (1 - cos) * width / 2. - sin * height / 2. + Tx], [-sin, cos, sin * width / 2. + (1 - cos) * height / 2. + Ty]]) 57 | image_rotate = cv2.warpAffine(image, M, (new_w, new_h), borderValue=(255, 255, 255)) 58 | landmarks = np.concatenate([landmarks, np.ones((landmarks.shape[0], 1))], axis=1) 59 | landmarks_rotate = np.dot(M, landmarks.T).T 60 | # return 61 | return self.crop(image_rotate, landmarks_rotate) 62 | '''crop''' 63 | def crop(self, image, landmarks): 64 | landmarks_top = np.min(landmarks[:, 1]) 65 | landmarks_bottom = np.max(landmarks[:, 1]) 66 | landmarks_left = np.min(landmarks[:, 0]) 67 | landmarks_right = np.max(landmarks[:, 0]) 68 | top = int(landmarks_top - 0.8 * (landmarks_bottom - landmarks_top)) 69 | bottom = int(landmarks_bottom + 0.3 * (landmarks_bottom - landmarks_top)) 70 | left = int(landmarks_left - 0.3 * (landmarks_right - landmarks_left)) 71 | right = int(landmarks_right + 0.3 * (landmarks_right - landmarks_left)) 72 | if bottom - top > right - left: 73 | left -= ((bottom - top) - (right - left)) // 2 74 | right = left + (bottom - top) 75 | else: 76 | top -= ((right - left) - (bottom - top)) // 2 77 | bottom = top + (right - left) 78 | image_crop = np.ones((bottom - top + 1, right - left + 1, 3), np.uint8) * 255 79 | h, w = image.shape[:2] 80 | left_white = max(0, -left) 81 | left = max(0, left) 82 | right = min(right, w-1) 83 | right_white = left_white + (right-left) 84 | top_white = max(0, -top) 85 | top = max(0, top) 86 | bottom = min(bottom, h-1) 87 | bottom_white = top_white + (bottom - top) 88 | image_crop[top_white:bottom_white+1, left_white:right_white+1] = image[top:bottom+1, left:right+1].copy() 89 | return image_crop -------------------------------------------------------------------------------- /pydrawing/modules/beautifiers/cartoonizeface/facesegmentor/__init__.py: -------------------------------------------------------------------------------- 1 | '''initialize''' 2 | from .facesegmentor import FaceSegmentor -------------------------------------------------------------------------------- /pydrawing/modules/beautifiers/cartoonizeface/facesegmentor/facesegmentor.py: -------------------------------------------------------------------------------- 1 | ''' 2 | Function: 3 | 人脸分割 4 | Author: 5 | Charles 6 | 微信公众号: 7 | Charles的皮卡丘 8 | ''' 9 | import cv2 10 | import torch 11 | import numpy as np 12 | import torch.nn as nn 13 | import torch.utils.model_zoo as model_zoo 14 | 15 | 16 | '''config''' 17 | SEGMENTOR_CFG = { 18 | 'type': 'ce2p', 19 | 'benchmark': True, 20 | 'num_classes': -1, 21 | 'align_corners': False, 22 | 'is_multi_gpus': True, 23 | 'distributed': {'is_on': True, 'backend': 'nccl'}, 24 | 'norm_cfg': {'type': 'batchnorm2d', 'opts': {}}, 25 | 'act_cfg': {'type': 'leakyrelu', 'opts': {'negative_slope': 0.01, 'inplace': True}}, 26 | 'backbone': { 27 | 'type': 'resnet101', 28 | 'series': 'resnet', 29 | 'pretrained': False, 30 | 'outstride': 16, 31 | 'use_stem': True, 32 | 'selected_indices': (0, 1, 2, 3), 33 | }, 34 | 'ppm': { 35 | 'in_channels': 2048, 36 | 'out_channels': 512, 37 | 'pool_scales': [1, 2, 3, 6], 38 | }, 39 | 'epm': { 40 | 'in_channels_list': [256, 512, 1024], 41 | 'hidden_channels': 256, 42 | 'out_channels': 2 43 | }, 44 | 'shortcut': { 45 | 'in_channels': 256, 46 | 'out_channels': 48, 47 | }, 48 | 'decoder':{ 49 | 'stage1': { 50 | 'in_channels': 560, 51 | 'out_channels': 512, 52 | 'dropout': 0, 53 | }, 54 | 'stage2': { 55 | 'in_channels': 1280, 56 | 'out_channels': 512, 57 | 'dropout': 0.1 58 | }, 59 | }, 60 | } 61 | SEGMENTOR_CFG.update( 62 | { 63 | 'num_classes': 20, 64 | 'backbone': { 65 | 'type': 'resnet50', 66 | 'series': 'resnet', 67 | 'pretrained': True, 68 | 'outstride': 8, 69 | 'use_stem': True, 70 | 'selected_indices': (0, 1, 2, 3), 71 | } 72 | } 73 | ) 74 | 75 | 76 | '''FaceSegmentor''' 77 | class FaceSegmentor(nn.Module): 78 | def __init__(self, **kwargs): 79 | super(FaceSegmentor, self).__init__() 80 | try: 81 | from ssseg.modules.models.segmentors.ce2p import CE2P 82 | except: 83 | raise RuntimeError('Please run "pip install sssegmentation" to install "ssseg"') 84 | self.ce2p = CE2P(SEGMENTOR_CFG, mode='TEST') 85 | self.ce2p.load_state_dict(model_zoo.load_url('https://github.com/SegmentationBLWX/modelstore/releases/download/ssseg_ce2p/ce2p_resnet50os8_lip_train.pth', map_location='cpu')['model']) 86 | '''forward''' 87 | def forward(self, x): 88 | return self.ce2p(x) 89 | '''preprocess''' 90 | def preprocess(self, image): 91 | # Resize 92 | output_size = (473, 473) 93 | if image.shape[0] > image.shape[1]: 94 | dsize = min(output_size), max(output_size) 95 | else: 96 | dsize = max(output_size), min(output_size) 97 | image = cv2.resize(image, dsize=dsize, interpolation=cv2.INTER_LINEAR) 98 | # Normalize 99 | mean, std = np.array([123.675, 116.28, 103.53]), np.array([58.395, 57.12, 57.375]) 100 | image = image.astype(np.float32) 101 | mean = np.float64(mean.reshape(1, -1)) 102 | stdinv = 1 / np.float64(std.reshape(1, -1)) 103 | cv2.cvtColor(image, cv2.COLOR_BGR2RGB, image) 104 | cv2.subtract(image, mean, image) 105 | cv2.multiply(image, stdinv, image) 106 | # ToTensor 107 | image = torch.from_numpy((image.transpose((2, 0, 1))).astype(np.float32)) 108 | # Return 109 | return image.unsqueeze(0) 110 | '''get face mask''' 111 | def getfacemask(self, mask): 112 | output_mask = np.zeros(mask.shape[:2]) 113 | face_idxs = [2, 13] 114 | for idx in face_idxs: 115 | output_mask[mask == idx] = 255 116 | return output_mask -------------------------------------------------------------------------------- /pydrawing/modules/beautifiers/characterize/__init__.py: -------------------------------------------------------------------------------- 1 | '''initialize''' 2 | from .characterize import CharacterizeBeautifier -------------------------------------------------------------------------------- /pydrawing/modules/beautifiers/characterize/characterize.py: -------------------------------------------------------------------------------- 1 | ''' 2 | Function: 3 | 视频转字符画 4 | Author: 5 | Charles 6 | 微信公众号: 7 | Charles的皮卡丘 8 | ''' 9 | import cv2 10 | import numpy as np 11 | from ..base import BaseBeautifier 12 | from PIL import Image, ImageFont, ImageDraw 13 | 14 | 15 | '''视频转字符画''' 16 | class CharacterizeBeautifier(BaseBeautifier): 17 | CHARS = "$@B%8&WM#*oahkbdpqwmZO0QLCJUYXzcvunxrjft/\|()1{}[]?-_+~<>i!lI;:,\"^`'. " 18 | def __init__(self, **kwargs): 19 | super(CharacterizeBeautifier, self).__init__(**kwargs) 20 | '''迭代图片''' 21 | def iterimage(self, image): 22 | image = Image.fromarray(cv2.cvtColor(image, cv2.COLOR_BGR2RGB)) 23 | # 每个字符的大小 24 | font = ImageFont.load_default().font 25 | font_w, font_h = font.getsize(self.CHARS[1]) 26 | # 输入图像变成可整除字符大小 27 | image = image.resize((font_w * (image.width // font_w), font_h * (image.height // font_h)), Image.NEAREST) 28 | # 原始大小 29 | h_ori, w_ori = image.height, image.width 30 | # resize 31 | image = image.resize((w_ori // font_w, h_ori // font_h), Image.NEAREST) 32 | h, w = image.height, image.width 33 | # 图像RGB转字符 34 | txts, colors = '', [] 35 | for i in range(h): 36 | for j in range(w): 37 | pixel = image.getpixel((j, i)) 38 | colors.append(pixel[:3]) 39 | txts += self.rgb2char(*pixel) 40 | image = Image.new('RGB', (w_ori, h_ori), (255, 255, 255)) 41 | draw = ImageDraw.Draw(image) 42 | x = y = 0 43 | for j in range(len(txts)): 44 | if x == w_ori: x, y = 0, y + font_h 45 | draw.text((x, y), txts[j], font=font, fill=colors[j]) 46 | x += font_w 47 | return cv2.cvtColor(np.asarray(image), cv2.COLOR_RGB2BGR) 48 | '''RGB转字符''' 49 | def rgb2char(self, r, g, b, alpha=256): 50 | if alpha == 0: return '' 51 | gray = int(0.2126 * r + 0.7152 * g + 0.0722 * b) 52 | return self.CHARS[gray % len(self.CHARS)] -------------------------------------------------------------------------------- /pydrawing/modules/beautifiers/douyineffect/__init__.py: -------------------------------------------------------------------------------- 1 | '''initialize''' 2 | from .douyineffect import DouyinEffectBeautifier -------------------------------------------------------------------------------- /pydrawing/modules/beautifiers/douyineffect/douyineffect.py: -------------------------------------------------------------------------------- 1 | ''' 2 | Function: 3 | 图像抖音特效画 4 | Author: 5 | Charles 6 | 微信公众号: 7 | Charles的皮卡丘 8 | ''' 9 | import cv2 10 | import copy 11 | import numpy as np 12 | from PIL import Image 13 | from ..base import BaseBeautifier 14 | 15 | 16 | '''图像抖音特效画''' 17 | class DouyinEffectBeautifier(BaseBeautifier): 18 | def __init__(self, **kwargs): 19 | super(DouyinEffectBeautifier, self).__init__(**kwargs) 20 | '''迭代图片''' 21 | def iterimage(self, image): 22 | image = cv2.cvtColor(image, cv2.COLOR_BGR2RGBA) 23 | # 提取R 24 | image_arr_r = copy.deepcopy(image) 25 | image_arr_r[:, :, 1:3] = 0 26 | # 提取GB 27 | image_arr_gb = copy.deepcopy(image) 28 | image_arr_gb[:, :, 0] = 0 29 | # 创建画布把图片错开放 30 | image_r = Image.fromarray(image_arr_r).convert('RGBA') 31 | image_gb = Image.fromarray(image_arr_gb).convert('RGBA') 32 | canvas_r = Image.new('RGB', (image.shape[1], image.shape[0]), color=(0, 0, 0)) 33 | canvas_gb = Image.new('RGB', (image.shape[1], image.shape[0]), color=(0, 0, 0)) 34 | canvas_r.paste(image_r, (6, 6), image_r) 35 | canvas_gb.paste(image_gb, (0, 0), image_gb) 36 | output_image = np.array(canvas_gb) + np.array(canvas_r) 37 | output_image = cv2.cvtColor(output_image, cv2.COLOR_RGB2BGR) 38 | # 返回结果 39 | return output_image -------------------------------------------------------------------------------- /pydrawing/modules/beautifiers/fastneuralstyletransfer/__init__.py: -------------------------------------------------------------------------------- 1 | '''initialize''' 2 | from .fastneuralstyletransfer import FastNeuralStyleTransferBeautifier -------------------------------------------------------------------------------- /pydrawing/modules/beautifiers/fastneuralstyletransfer/fastneuralstyletransfer.py: -------------------------------------------------------------------------------- 1 | ''' 2 | Function: 3 | 复现论文"Perceptual Losses for Real-Time Style Transfer and Super-Resolution" 4 | Author: 5 | Charles 6 | 微信公众号: 7 | Charles的皮卡丘 8 | ''' 9 | import cv2 10 | import numpy as np 11 | from ..base import BaseBeautifier 12 | try: 13 | import torch 14 | import torch.nn as nn 15 | import torch.nn.functional as F 16 | import torchvision.models as models 17 | import torch.utils.model_zoo as model_zoo 18 | import torchvision.transforms as transforms 19 | except: 20 | print('[Warning]: Pytorch and torchvision have not be installed, "fastneuralstyletransfer" will be not available.') 21 | 22 | 23 | '''ConvBlock''' 24 | class ConvBlock(nn.Module): 25 | def __init__(self, in_channels, out_channels, kernel_size, stride=1, upsample=False, normalize=True, relu=True): 26 | super(ConvBlock, self).__init__() 27 | self.upsample = upsample 28 | self.block = nn.Sequential( 29 | nn.ReflectionPad2d(kernel_size // 2), nn.Conv2d(in_channels, out_channels, kernel_size, stride) 30 | ) 31 | self.norm = nn.InstanceNorm2d(out_channels, affine=True) if normalize else None 32 | self.relu = relu 33 | '''forward''' 34 | def forward(self, x): 35 | if self.upsample: x = F.interpolate(x, scale_factor=2) 36 | x = self.block(x) 37 | if self.norm is not None: x = self.norm(x) 38 | if self.relu: x = F.relu(x) 39 | return x 40 | 41 | 42 | '''ResidualBlock''' 43 | class ResidualBlock(nn.Module): 44 | def __init__(self, channels): 45 | super(ResidualBlock, self).__init__() 46 | self.block = nn.Sequential( 47 | ConvBlock(channels, channels, kernel_size=3, stride=1, normalize=True, relu=True), 48 | ConvBlock(channels, channels, kernel_size=3, stride=1, normalize=True, relu=False), 49 | ) 50 | '''forward''' 51 | def forward(self, x): 52 | return self.block(x) + x 53 | 54 | 55 | '''TransformerNet, 模型修改自: https://github.com/eriklindernoren/Fast-Neural-Style-Transfer''' 56 | class TransformerNet(nn.Module): 57 | def __init__(self): 58 | super(TransformerNet, self).__init__() 59 | self.model = nn.Sequential( 60 | ConvBlock(3, 32, kernel_size=9, stride=1), 61 | ConvBlock(32, 64, kernel_size=3, stride=2), 62 | ConvBlock(64, 128, kernel_size=3, stride=2), 63 | ResidualBlock(128), 64 | ResidualBlock(128), 65 | ResidualBlock(128), 66 | ResidualBlock(128), 67 | ResidualBlock(128), 68 | ConvBlock(128, 64, kernel_size=3, upsample=True), 69 | ConvBlock(64, 32, kernel_size=3, upsample=True), 70 | ConvBlock(32, 3, kernel_size=9, stride=1, normalize=False, relu=False), 71 | ) 72 | '''forward''' 73 | def forward(self, x): 74 | return self.model(x) 75 | 76 | 77 | '''复现论文"Perceptual Losses for Real-Time Style Transfer and Super-Resolution"''' 78 | class FastNeuralStyleTransferBeautifier(BaseBeautifier): 79 | def __init__(self, style='starrynight', use_cuda=True, **kwargs): 80 | super(FastNeuralStyleTransferBeautifier, self).__init__(**kwargs) 81 | self.model_urls = { 82 | 'cuphead': 'https://github.com/CharlesPikachu/pydrawing/releases/download/checkpoints/fastneuralstyletransfer_cuphead.pth', 83 | 'mosaic': 'https://github.com/CharlesPikachu/pydrawing/releases/download/checkpoints/fastneuralstyletransfer_mosaic.pth', 84 | 'starrynight': 'https://github.com/CharlesPikachu/pydrawing/releases/download/checkpoints/fastneuralstyletransfer_starrynight.pth', 85 | } 86 | self.mean = np.array([0.485, 0.456, 0.406]) 87 | self.std = np.array([0.229, 0.224, 0.225]) 88 | self.preprocess = transforms.Compose([transforms.ToTensor(), transforms.Normalize(self.mean, self.std)]) 89 | assert style in self.model_urls 90 | self.style = style 91 | self.use_cuda = use_cuda 92 | self.transformer = TransformerNet() 93 | self.transformer.load_state_dict(model_zoo.load_url(self.model_urls[style])) 94 | self.transformer.eval() 95 | if torch.cuda.is_available() and self.use_cuda: self.transformer = self.transformer.cuda() 96 | '''迭代图片''' 97 | def iterimage(self, image): 98 | image = cv2.cvtColor(image, cv2.COLOR_BGR2RGB) 99 | input_image = self.preprocess(image).unsqueeze(0) 100 | if torch.cuda.is_available() and self.use_cuda: 101 | input_image = input_image.cuda() 102 | with torch.no_grad(): 103 | output_image = self.transformer(input_image)[0] 104 | output_image = output_image.data.cpu().float() 105 | for c in range(3): 106 | output_image[c, :].mul_(self.std[c]).add_(self.mean[c]) 107 | output_image = output_image.mul(255).add_(0.5).clamp_(0, 255).permute(1, 2, 0).to('cpu', torch.uint8).numpy() 108 | output_image = cv2.cvtColor(output_image, cv2.COLOR_RGB2BGR) 109 | output_image = cv2.resize(output_image, (image.shape[1], image.shape[0])) 110 | return output_image -------------------------------------------------------------------------------- /pydrawing/modules/beautifiers/geneticfitting/__init__.py: -------------------------------------------------------------------------------- 1 | '''initialize''' 2 | from .geneticfittingcircle import GeneticFittingCircleBeautifier 3 | from .geneticfittingpolygon import GeneticFittingPolygonBeautifier -------------------------------------------------------------------------------- /pydrawing/modules/beautifiers/geneticfitting/geneticfittingcircle.py: -------------------------------------------------------------------------------- 1 | ''' 2 | Function: 3 | 利用遗传算法画画 4 | Author: 5 | Charles 6 | 微信公众号: 7 | Charles的皮卡丘 8 | ''' 9 | import os 10 | import cv2 11 | import copy 12 | import random 13 | import numpy as np 14 | from ...utils import checkdir 15 | from PIL import Image, ImageDraw 16 | from ..base import BaseBeautifier 17 | 18 | 19 | '''圆形''' 20 | class Circle(): 21 | def __init__(self, radius_range=50, radius_shift_range=50, center_shift_range=50, color_shift_range=50, target_image=None): 22 | # 属性 23 | self.radius_range = radius_range 24 | self.radius_shift_range = radius_shift_range 25 | self.center_shift_range = center_shift_range 26 | self.color_shift_range = color_shift_range 27 | self.target_image = target_image 28 | # 中心点 29 | image_width, image_height = target_image.size[:2] 30 | x, y = random.randint(0, int(image_width)), random.randint(0, int(image_height)) 31 | self.image_width, self.image_height = image_width, image_height 32 | self.center = (x, y) 33 | # 半径 34 | self.radius = random.randint(1, radius_range) 35 | # 颜色 36 | r, g, b = np.asarray(target_image)[min(y, image_height-1), min(x, image_width-1)] 37 | self.color = (int(r), int(g), int(b), random.randint(0, 256)) 38 | '''变异''' 39 | def mutate(self): 40 | mutations = ['center', 'radius', 'color', 'reset'] 41 | mutation_type = random.choice(mutations) 42 | # 随机中心的偏移 43 | if mutation_type == 'center': 44 | x_shift = int(random.randint(-self.center_shift_range, self.center_shift_range) * random.random()) 45 | y_shift = int(random.randint(-self.center_shift_range, self.center_shift_range) * random.random()) 46 | x = min(max(0, x_shift + self.center[0]), self.image_width) 47 | y = min(max(0, y_shift + self.center[1]), self.image_height) 48 | self.center = (x, y) 49 | # 随机改变半径 50 | elif mutation_type == 'radius': 51 | self.radius += int(random.randint(-self.radius_shift_range, self.radius_shift_range) * random.random()) 52 | # 随机改变颜色 53 | elif mutation_type == 'color': 54 | self.color = tuple(c + int(random.randint(-self.color_shift_range, self.color_shift_range) * random.random()) for c in self.color) 55 | self.color = tuple(min(max(c, 0), 255) for c in self.color) 56 | # 重置 57 | else: 58 | new_circle = Circle( 59 | radius_range=self.radius_range, 60 | radius_shift_range=self.radius_shift_range, 61 | center_shift_range=self.center_shift_range, 62 | color_shift_range=self.color_shift_range, 63 | target_image=self.target_image, 64 | ) 65 | self.center = new_circle.center 66 | self.radius = new_circle.radius 67 | self.color = new_circle.color 68 | 69 | 70 | '''利用遗传算法画画''' 71 | class GeneticFittingCircleBeautifier(BaseBeautifier): 72 | def __init__(self, init_cfg=None, cache_dir='cache', save_cache=True, **kwargs): 73 | super(GeneticFittingCircleBeautifier, self).__init__(**kwargs) 74 | if init_cfg is None: 75 | init_cfg = { 76 | 'num_populations': 10, 77 | 'init_num_circles': 1, 78 | 'num_generations': 1e5, 79 | 'print_interval': 1, 80 | 'mutation_rate': 0.1, 81 | 'selection_rate': 0.5, 82 | 'crossover_rate': 0.5, 83 | 'circle_cfg': {'radius_range': 50, 'radius_shift_range': 50, 'center_shift_range': 50, 'color_shift_range': 50}, 84 | } 85 | self.init_cfg = init_cfg 86 | self.cache_dir = cache_dir 87 | self.save_cache = save_cache 88 | '''迭代图片''' 89 | def iterimage(self, image): 90 | image = Image.fromarray(cv2.cvtColor(image, cv2.COLOR_BGR2RGB)) 91 | # 初始化 92 | populations = [] 93 | for _ in range(self.init_cfg['num_populations']): 94 | population = [] 95 | for _ in range(self.init_cfg['init_num_circles']): 96 | population.append(Circle( 97 | target_image=image, 98 | **self.init_cfg['circle_cfg'] 99 | )) 100 | populations.append(population) 101 | # 迭代 102 | mutation_rate = self.init_cfg['mutation_rate'] 103 | for g in range(1, int(self.init_cfg['num_generations']+1)): 104 | fitnesses = [] 105 | for idx, population in enumerate(copy.deepcopy(populations)): 106 | fitness_ori = self.calcfitnesses([population], image)[0] 107 | fitness = 0 108 | while fitness_ori > fitness: 109 | population_new = population + [Circle(target_image=image, **self.init_cfg['circle_cfg'])] 110 | fitness = self.calcfitnesses([population_new], image)[0] 111 | populations[idx] = population_new 112 | fitnesses.append(fitness) 113 | if g % self.init_cfg['print_interval'] == 0: 114 | if self.save_cache: 115 | population = populations[np.argmax(fitnesses)] 116 | output_image = self.draw(population, image) 117 | checkdir(self.cache_dir) 118 | output_image.save(os.path.join(self.cache_dir, f'cache_g{g}.png')) 119 | self.logger_handle.info(f'Generation: {g}, FITNESS: {max(fitnesses)}') 120 | num_populations = len(populations) 121 | # --自然选择 122 | populations = self.select(image, fitnesses, populations) 123 | # --交叉 124 | populations = self.crossover(populations, num_populations) 125 | # --变异 126 | populations = self.mutate(image, populations, mutation_rate) 127 | # 选择最优解返回 128 | population = populations[np.argmax(fitnesses)] 129 | output_image = self.draw(population, image) 130 | return cv2.cvtColor(np.asarray(output_image), cv2.COLOR_RGB2BGR) 131 | '''自然选择''' 132 | def select(self, image, fitnesses, populations): 133 | sorted_idx = np.argsort(fitnesses)[::-1] 134 | selected_populations = [] 135 | selection_rate = self.init_cfg['selection_rate'] 136 | for idx in range(int(len(populations) * selection_rate)): 137 | selected_idx = int(sorted_idx[idx]) 138 | selected_populations.append(populations[selected_idx]) 139 | return selected_populations 140 | '''交叉''' 141 | def crossover(self, populations, num_populations): 142 | indices = list(range(len(populations))) 143 | while len(populations) < num_populations: 144 | idx1 = random.choice(indices) 145 | idx2 = random.choice(indices) 146 | population1 = copy.deepcopy(populations[idx1]) 147 | population2 = copy.deepcopy(populations[idx2]) 148 | for circle_idx in range(len(population1)): 149 | if self.init_cfg['crossover_rate'] > random.random(): 150 | population1[circle_idx] = population2[circle_idx] 151 | populations.append(population1) 152 | return populations 153 | '''变异''' 154 | def mutate(self, target_image, populations, mutation_rate): 155 | populations_new = copy.deepcopy(populations) 156 | for idx, population in enumerate(populations): 157 | fitness_ori = self.calcfitnesses([population], target_image)[0] 158 | for circle in population: 159 | if mutation_rate > random.random(): 160 | circle.mutate() 161 | fitness = self.calcfitnesses([population], target_image)[0] 162 | if fitness > fitness_ori: populations_new[idx] = population 163 | return populations_new 164 | '''计算适应度''' 165 | def calcfitnesses(self, populations, target_image): 166 | fitnesses = [] 167 | for idx in range(len(populations)): 168 | image = self.draw(populations[idx], target_image) 169 | fitnesses.append(self.calcsimilarity(image, target_image)) 170 | return fitnesses 171 | '''画图''' 172 | def draw(self, population, target_image): 173 | image = Image.new('RGB', target_image.size, "#FFFFFF") 174 | for circle in population: 175 | item = Image.new('RGB', target_image.size, "#000000") 176 | mask = Image.new('L', target_image.size, 255) 177 | draw = ImageDraw.Draw(item) 178 | draw.ellipse((circle.center[0]-circle.radius, circle.center[1]-circle.radius, circle.center[0]+circle.radius, circle.center[1]+circle.radius), fill=circle.color) 179 | draw = ImageDraw.Draw(mask) 180 | draw.ellipse((circle.center[0]-circle.radius, circle.center[1]-circle.radius, circle.center[0]+circle.radius, circle.center[1]+circle.radius), fill=128) 181 | image = Image.composite(image, item, mask) 182 | return image 183 | '''计算两幅图之间的相似度''' 184 | def calcsimilarity(self, image, target_image): 185 | image = np.asarray(image) / 255. 186 | target_image = np.asarray(target_image) / 255. 187 | similarity = 1 - abs(image - target_image).mean() 188 | return similarity -------------------------------------------------------------------------------- /pydrawing/modules/beautifiers/geneticfitting/geneticfittingpolygon.py: -------------------------------------------------------------------------------- 1 | ''' 2 | Function: 3 | 利用遗传算法画画 4 | Author: 5 | Charles 6 | 微信公众号: 7 | Charles的皮卡丘 8 | ''' 9 | import os 10 | import cv2 11 | import copy 12 | import random 13 | import numpy as np 14 | from ...utils import checkdir 15 | from PIL import Image, ImageDraw 16 | from ..base import BaseBeautifier 17 | 18 | 19 | '''多边形''' 20 | class Polygon(): 21 | def __init__(self, num_points=3, size=50, shift_range=50, point_range=50, color_range=50, target_image=None, **kwargs): 22 | # set attrs 23 | self.size = size 24 | self.num_points = num_points 25 | self.target_image = target_image 26 | self.shift_range = shift_range 27 | self.point_range = point_range 28 | self.color_range = color_range 29 | # 点 30 | image_width, image_height = target_image.size[:2] 31 | x, y = random.randint(0, int(image_width)), random.randint(0, int(image_height)) 32 | self.points = [] 33 | for _ in range(self.num_points): 34 | self.points.append(((y + random.randint(-size, size), x + random.randint(-size, size)))) 35 | # 颜色 36 | point = random.choice(self.points) 37 | r, g, b = np.asarray(target_image)[min(point[0], image_height-1), min(point[1], image_width-1)] 38 | self.color = (int(r), int(g), int(b), random.randint(0, 256)) 39 | '''变异''' 40 | def mutate(self): 41 | mutations = ['shift', 'point', 'color', 'reset'] 42 | mutation_type = random.choice(mutations) 43 | # 整体偏移 44 | if mutation_type == 'shift': 45 | x_shift = int(random.randint(-self.shift_range, self.shift_range) * random.random()) 46 | y_shift = int(random.randint(-self.shift_range, self.shift_range) * random.random()) 47 | self.points = [(x + x_shift, y + y_shift) for x, y in self.points] 48 | # 随机改变一个点 49 | elif mutation_type == 'point': 50 | index = random.choice(list(range(len(self.points)))) 51 | self.points[index] = ( 52 | self.points[index][0] + int(random.randint(-self.point_range, self.point_range) * random.random()), 53 | self.points[index][1] + int(random.randint(-self.point_range, self.point_range) * random.random()), 54 | ) 55 | # 随机改变颜色 56 | elif mutation_type == 'color': 57 | self.color = tuple(c + int(random.randint(-self.color_range, self.color_range) * random.random()) for c in self.color) 58 | self.color = tuple(min(max(c, 0), 255) for c in self.color) 59 | # 重置 60 | else: 61 | new_polygon = Polygon( 62 | num_points=max(self.num_points + random.choice([-1, 0, 1]), 3), 63 | size=self.size, 64 | shift_range=self.shift_range, 65 | point_range=self.point_range, 66 | color_range=self.color_range, 67 | target_image=self.target_image 68 | ) 69 | self.points = new_polygon.points 70 | self.color = new_polygon.color 71 | 72 | 73 | '''利用遗传算法画画''' 74 | class GeneticFittingPolygonBeautifier(BaseBeautifier): 75 | def __init__(self, init_cfg=None, cache_dir='cache', save_cache=True, **kwargs): 76 | super(GeneticFittingPolygonBeautifier, self).__init__(**kwargs) 77 | if init_cfg is None: 78 | init_cfg = { 79 | 'num_populations': 10, 80 | 'num_points_list': list(range(3, 40)), 81 | 'init_num_polygons': 1, 82 | 'num_generations': 1e5, 83 | 'print_interval': 1, 84 | 'mutation_rate': 0.1, 85 | 'selection_rate': 0.5, 86 | 'crossover_rate': 0.5, 87 | 'polygon_cfg': {'size': 50, 'shift_range': 50, 'point_range': 50, 'color_range': 50}, 88 | } 89 | self.init_cfg = init_cfg 90 | self.cache_dir = cache_dir 91 | self.save_cache = save_cache 92 | '''迭代图片''' 93 | def iterimage(self, image): 94 | image = Image.fromarray(cv2.cvtColor(image, cv2.COLOR_BGR2RGB)) 95 | # 初始化 96 | populations = [] 97 | for _ in range(self.init_cfg['num_populations']): 98 | population = [] 99 | for _ in range(self.init_cfg['init_num_polygons']): 100 | population.append(Polygon( 101 | target_image=image, 102 | num_points=random.choice(self.init_cfg['num_points_list']), 103 | **self.init_cfg['polygon_cfg'] 104 | )) 105 | populations.append(population) 106 | # 迭代 107 | mutation_rate = self.init_cfg['mutation_rate'] 108 | for g in range(1, int(self.init_cfg['num_generations']+1)): 109 | fitnesses = [] 110 | for idx, population in enumerate(copy.deepcopy(populations)): 111 | fitness_ori = self.calcfitnesses([population], image)[0] 112 | fitness = 0 113 | while fitness_ori > fitness: 114 | population_new = population + [Polygon(target_image=image, num_points=random.choice(self.init_cfg['num_points_list']), **self.init_cfg['polygon_cfg'])] 115 | fitness = self.calcfitnesses([population_new], image)[0] 116 | populations[idx] = population_new 117 | fitnesses.append(fitness) 118 | if g % self.init_cfg['print_interval'] == 0: 119 | if self.save_cache: 120 | population = populations[np.argmax(fitnesses)] 121 | output_image = self.draw(population, image) 122 | checkdir(self.cache_dir) 123 | output_image.save(os.path.join(self.cache_dir, f'cache_g{g}.png')) 124 | self.logger_handle.info(f'Generation: {g}, FITNESS: {max(fitnesses)}') 125 | num_populations = len(populations) 126 | # --自然选择 127 | populations = self.select(image, fitnesses, populations) 128 | # --交叉 129 | populations = self.crossover(populations, num_populations) 130 | # --变异 131 | populations = self.mutate(image, populations, mutation_rate) 132 | # 选择最优解返回 133 | population = populations[np.argmax(fitnesses)] 134 | output_image = self.draw(population, image) 135 | return cv2.cvtColor(np.asarray(output_image), cv2.COLOR_RGB2BGR) 136 | '''自然选择''' 137 | def select(self, image, fitnesses, populations): 138 | sorted_idx = np.argsort(fitnesses)[::-1] 139 | selected_populations = [] 140 | selection_rate = self.init_cfg['selection_rate'] 141 | for idx in range(int(len(populations) * selection_rate)): 142 | selected_idx = int(sorted_idx[idx]) 143 | selected_populations.append(populations[selected_idx]) 144 | return selected_populations 145 | '''交叉''' 146 | def crossover(self, populations, num_populations): 147 | indices = list(range(len(populations))) 148 | while len(populations) < num_populations: 149 | idx1 = random.choice(indices) 150 | idx2 = random.choice(indices) 151 | population1 = copy.deepcopy(populations[idx1]) 152 | population2 = copy.deepcopy(populations[idx2]) 153 | for polygon_idx in range(len(population1)): 154 | if self.init_cfg['crossover_rate'] > random.random(): 155 | population1[polygon_idx] = population2[polygon_idx] 156 | populations.append(population1) 157 | return populations 158 | '''变异''' 159 | def mutate(self, target_image, populations, mutation_rate): 160 | populations_new = copy.deepcopy(populations) 161 | for idx, population in enumerate(populations): 162 | fitness_ori = self.calcfitnesses([population], target_image)[0] 163 | for polygon in population: 164 | if mutation_rate > random.random(): 165 | polygon.mutate() 166 | fitness = self.calcfitnesses([population], target_image)[0] 167 | if fitness > fitness_ori: populations_new[idx] = population 168 | return populations_new 169 | '''计算适应度''' 170 | def calcfitnesses(self, populations, target_image): 171 | fitnesses = [] 172 | for idx in range(len(populations)): 173 | image = self.draw(populations[idx], target_image) 174 | fitnesses.append(self.calcsimilarity(image, target_image)) 175 | return fitnesses 176 | '''画图''' 177 | def draw(self, population, target_image): 178 | image = Image.new('RGB', target_image.size, "#FFFFFF") 179 | for polygon in population: 180 | item = Image.new('RGB', target_image.size, "#000000") 181 | mask = Image.new('L', target_image.size, 255) 182 | draw = ImageDraw.Draw(item) 183 | draw.polygon(polygon.points, polygon.color) 184 | draw = ImageDraw.Draw(mask) 185 | draw.polygon(polygon.points, fill=128) 186 | image = Image.composite(image, item, mask) 187 | return image 188 | '''计算两幅图之间的相似度''' 189 | def calcsimilarity(self, image, target_image): 190 | image = np.asarray(image) / 255. 191 | target_image = np.asarray(target_image) / 255. 192 | similarity = 1 - abs(image - target_image).mean() 193 | return similarity -------------------------------------------------------------------------------- /pydrawing/modules/beautifiers/glitch/__init__.py: -------------------------------------------------------------------------------- 1 | '''initialize''' 2 | from .glitch import GlitchBeautifier -------------------------------------------------------------------------------- /pydrawing/modules/beautifiers/glitch/glitch.py: -------------------------------------------------------------------------------- 1 | ''' 2 | Function: 3 | 信号故障的效果 4 | Author: 5 | Charles 6 | 微信公众号: 7 | Charles的皮卡丘 8 | ''' 9 | import os 10 | import random 11 | from ...utils import checkdir 12 | from ..base import BaseBeautifier 13 | 14 | 15 | '''信号故障的效果''' 16 | class GlitchBeautifier(BaseBeautifier): 17 | def __init__(self, header_size=200, intensity=0.1, block_size=100, **kwargs): 18 | super(GlitchBeautifier, self).__init__(**kwargs) 19 | self.header_size, self.intensity, self.block_size = header_size, intensity, block_size 20 | '''处理文件''' 21 | def process(self, filepath): 22 | checkdir(self.savedir) 23 | ext = filepath.split('.')[-1] 24 | assert ext.lower() in ['mp4', 'avi'] 25 | with open(filepath, 'rb') as fp_in: 26 | with open(os.path.join(self.savedir, f'{self.savename}.{ext}'), 'wb') as fp_out: 27 | fp_out.write(fp_in.read(self.header_size)) 28 | while True: 29 | block_data = fp_in.read(self.block_size) 30 | if not block_data: break 31 | if random.random() < self.intensity / 100: block_data = os.urandom(self.block_size) 32 | fp_out.write(block_data) 33 | self.logger_handle.info(f'Video is saved into {self.savename}.{ext}') -------------------------------------------------------------------------------- /pydrawing/modules/beautifiers/nostalgicstyle/__init__.py: -------------------------------------------------------------------------------- 1 | '''initialize''' 2 | from .nostalgicstyle import NostalgicstyleBeautifier -------------------------------------------------------------------------------- /pydrawing/modules/beautifiers/nostalgicstyle/nostalgicstyle.py: -------------------------------------------------------------------------------- 1 | ''' 2 | Function: 3 | 照片怀旧风格 4 | Author: 5 | Charles 6 | 微信公众号: 7 | Charles的皮卡丘 8 | ''' 9 | import numpy as np 10 | from ..base import BaseBeautifier 11 | 12 | 13 | '''照片怀旧风格''' 14 | class NostalgicstyleBeautifier(BaseBeautifier): 15 | def __init__(self, **kwargs): 16 | super(NostalgicstyleBeautifier, self).__init__(**kwargs) 17 | '''迭代图片''' 18 | def iterimage(self, image): 19 | image = image.astype(np.float32) 20 | image_processed = image.copy() 21 | image_processed[..., 0] = image[..., 2] * 0.272 + image[..., 1] * 0.534 + image[..., 0] * 0.131 22 | image_processed[..., 1] = image[..., 2] * 0.349 + image[..., 1] * 0.686 + image[..., 0] * 0.168 23 | image_processed[..., 2] = image[..., 2] * 0.393 + image[..., 1] * 0.769 + image[..., 0] * 0.189 24 | image_processed[image_processed > 255.0] = 255.0 25 | return image_processed -------------------------------------------------------------------------------- /pydrawing/modules/beautifiers/noteprocessor/__init__.py: -------------------------------------------------------------------------------- 1 | '''initialize''' 2 | from .noteprocessor import NoteprocessorBeautifier -------------------------------------------------------------------------------- /pydrawing/modules/beautifiers/noteprocessor/noteprocessor.py: -------------------------------------------------------------------------------- 1 | ''' 2 | Function: 3 | 手写笔记处理 4 | Author: 5 | Charles 6 | 微信公众号: 7 | Charles的皮卡丘 8 | ''' 9 | import os 10 | import cv2 11 | import numpy as np 12 | from PIL import Image 13 | from ..base import BaseBeautifier 14 | from scipy.cluster.vq import kmeans, vq 15 | 16 | 17 | '''手写笔记处理''' 18 | class NoteprocessorBeautifier(BaseBeautifier): 19 | def __init__(self, value_threshold=0.25, sat_threshold=0.20, num_colors=8, sample_fraction=0.05, white_bg=False, saturate=True, **kwargs): 20 | super(NoteprocessorBeautifier, self).__init__(**kwargs) 21 | self.num_colors = num_colors 22 | self.sample_fraction = sample_fraction 23 | self.value_threshold = value_threshold 24 | self.sat_threshold = sat_threshold 25 | self.white_bg = white_bg 26 | self.saturate = saturate 27 | '''迭代图片''' 28 | def iterimage(self, image): 29 | image = cv2.cvtColor(image, cv2.COLOR_BGR2RGB) 30 | sampled_pixels = self.getsampledpixels(image, self.sample_fraction) 31 | palette = self.getpalette(sampled_pixels) 32 | labels = self.applypalette(image, palette) 33 | if self.saturate: 34 | palette = palette.astype(np.float32) 35 | pmin = palette.min() 36 | pmax = palette.max() 37 | palette = 255 * (palette - pmin) / (pmax - pmin) 38 | palette = palette.astype(np.uint8) 39 | if self.white_bg: 40 | palette = palette.copy() 41 | palette[0] = (255, 255, 255) 42 | image_processed = Image.fromarray(labels, 'P') 43 | image_processed.putpalette(palette.flatten()) 44 | image_processed.save('tmp.png', dpi=(300, 300)) 45 | image_processed = cv2.imread('tmp.png') 46 | os.remove('tmp.png') 47 | return image_processed 48 | '''将调色板应用到给定的图像, 第一步是将所有背景像素设置为背景颜色, 之后是使用最邻近匹配将每个前景色映射到调色板中最接近的一个''' 49 | def applypalette(self, image, palette): 50 | bg_color = palette[0] 51 | fg_mask = self.getfgmask(bg_color, image) 52 | orig_shape = image.shape 53 | pixels = image.reshape((-1, 3)) 54 | fg_mask = fg_mask.flatten() 55 | num_pixels = pixels.shape[0] 56 | labels = np.zeros(num_pixels, dtype=np.uint8) 57 | labels[fg_mask], _ = vq(pixels[fg_mask], palette) 58 | return labels.reshape(orig_shape[:-1]) 59 | '''选取图像中固定百分比的像素, 以随机顺序返回''' 60 | def getsampledpixels(self, image, sample_fraction): 61 | pixels = image.reshape((-1, 3)) 62 | num_pixels = pixels.shape[0] 63 | num_samples = int(num_pixels * sample_fraction) 64 | idx = np.arange(num_pixels) 65 | np.random.shuffle(idx) 66 | return pixels[idx[:num_samples]] 67 | '''提取采样的RGB值集合的调色板, 调色板第一个条目始终是背景色, 其余的是通过运行K均值聚类从前景像素确定的''' 68 | def getpalette(self, samples, return_mask=False, kmeans_iter=40): 69 | bg_color = self.getbgcolor(samples, 6) 70 | fg_mask = self.getfgmask(bg_color, samples) 71 | centers, _ = kmeans(samples[fg_mask].astype(np.float32), self.num_colors-1, iter=kmeans_iter) 72 | palette = np.vstack((bg_color, centers)).astype(np.uint8) 73 | if not return_mask: return palette 74 | return palette, fg_mask 75 | '''通过与背景颜色进行比较来确定一组样本中的每个像素是否为前景, 如果像素的值或饱和度与背景的阈值不同, 则像素被分类为前景像素''' 76 | def getfgmask(self, bg_color, samples): 77 | s_bg, v_bg = self.rgbtosv(bg_color) 78 | s_samples, v_samples = self.rgbtosv(samples) 79 | s_diff = np.abs(s_bg - s_samples) 80 | v_diff = np.abs(v_bg - v_samples) 81 | return ((v_diff >= self.value_threshold) | (s_diff >= self.sat_threshold)) 82 | '''将RGB图像或RGB颜色数组转换为饱和度和数值, 每个都作为单独的32位浮点数组或值返回''' 83 | def rgbtosv(self, rgb): 84 | if not isinstance(rgb, np.ndarray): rgb = np.array(rgb) 85 | axis = len(rgb.shape) - 1 86 | cmax = rgb.max(axis=axis).astype(np.float32) 87 | cmin = rgb.min(axis=axis).astype(np.float32) 88 | delta = cmax - cmin 89 | saturation = delta.astype(np.float32) / cmax.astype(np.float32) 90 | saturation = np.where(cmax==0, 0, saturation) 91 | value = cmax / 255.0 92 | return saturation, value 93 | '''从图像或RGB颜色数组中获得背景颜色, 方法为通过将相似的颜色分组为相同的颜色并找到最常见的颜色''' 94 | def getbgcolor(self, image, bits_per_channel=6): 95 | assert image.shape[-1] == 3 96 | image_quantized = self.quantize(image, bits_per_channel).astype(int) 97 | image_packed = self.packrgb(image_quantized) 98 | unique, counts = np.unique(image_packed, return_counts=True) 99 | packed_mode = unique[counts.argmax()] 100 | return self.unpackrgb(packed_mode) 101 | '''减少给定图像中RGB三通道的位数''' 102 | def quantize(self, image, bits_per_channel=6): 103 | assert image.dtype == np.uint8 104 | shift = 8 - bits_per_channel 105 | halfbin = (1 << shift) >> 1 106 | return ((image.astype(int) >> shift) << shift) + halfbin 107 | '''将24位RGB三元组打包成一个整数, 参数rgb为元组或者数组''' 108 | def packrgb(self, rgb): 109 | orig_shape = None 110 | if isinstance(rgb, np.ndarray): 111 | assert rgb.shape[-1] == 3 112 | orig_shape = rgb.shape[:-1] 113 | else: 114 | assert len(rgb) == 3 115 | rgb = np.array(rgb) 116 | rgb = rgb.astype(int).reshape((-1, 3)) 117 | packed = (rgb[:, 0] << 16 | rgb[:, 1] << 8 | rgb[:, 2]) 118 | if orig_shape is None: return packed 119 | return packed.reshape(orig_shape) 120 | '''将一个整数或整数数组解压缩为一个或多个24位RGB值''' 121 | def unpackrgb(self, packed): 122 | orig_shape = None 123 | if isinstance(packed, np.ndarray): 124 | assert packed.dtype == int 125 | orig_shape = packed.shape 126 | packed = packed.reshape((-1, 1)) 127 | rgb = ((packed >> 16) & 0xff, (packed >> 8) & 0xff, (packed) & 0xff) 128 | if orig_shape is None: return rgb 129 | return np.hstack(rgb).reshape(orig_shape + (3,)) -------------------------------------------------------------------------------- /pydrawing/modules/beautifiers/oilpainting/__init__.py: -------------------------------------------------------------------------------- 1 | '''initialize''' 2 | from .oilpainting import OilpaintingBeautifier -------------------------------------------------------------------------------- /pydrawing/modules/beautifiers/oilpainting/oilpainting.py: -------------------------------------------------------------------------------- 1 | ''' 2 | Function: 3 | 照片油画化 4 | Author: 5 | Charles 6 | 微信公众号: 7 | Charles的皮卡丘 8 | ''' 9 | import cv2 10 | import random 11 | import numpy as np 12 | from scipy import ndimage 13 | from ..base import BaseBeautifier 14 | 15 | 16 | '''照片油画化''' 17 | class OilpaintingBeautifier(BaseBeautifier): 18 | def __init__(self, brush_width=5, palette=0, edge_operator='sobel', **kwargs): 19 | super(OilpaintingBeautifier, self).__init__(**kwargs) 20 | assert edge_operator in ['scharr', 'prewitt', 'sobel', 'roberts'] 21 | self.brush_width = brush_width 22 | self.palette = palette 23 | self.edge_operator = edge_operator 24 | '''迭代图片''' 25 | def iterimage(self, image): 26 | # 计算图像梯度 27 | r = 2 * int(image.shape[0] / 50) + 1 28 | gx, gy = self.getgradient(cv2.cvtColor(image, cv2.COLOR_BGR2GRAY), (r, r), self.edge_operator) 29 | gh = np.sqrt(np.sqrt(np.square(gx) + np.square(gy))) 30 | ga = (np.arctan2(gy, gx) / np.pi) * 180 + 90 31 | # 画油画的所有位置 32 | canvas = cv2.medianBlur(image, 11) 33 | order = self.getdraworder(image.shape[0], image.shape[1], scale=self.brush_width * 2) 34 | # 画椭圆 35 | colors = np.array(image, dtype=np.float) 36 | for i, (y, x) in enumerate(order): 37 | length = int(round(self.brush_width + self.brush_width * gh[y, x])) 38 | if self.palette != 0: 39 | color = np.array([round(colors[y, x][0] / self.palette) * self.palette + random.randint(-5, 5), \ 40 | round(colors[y, x][1] / self.palette) * self.palette + random.randint(-5, 5), \ 41 | round(colors[y, x][2] / self.palette) * self.palette + random.randint(-5, 5)], dtype=np.float) 42 | else: 43 | color = colors[y, x] 44 | cv2.ellipse(canvas, (x, y), (length, self.brush_width), ga[y, x], 0, 360, color, -1, cv2.LINE_AA) 45 | # 返回结果 46 | return canvas 47 | '''画油画的所有位置''' 48 | def getdraworder(self, h, w, scale): 49 | order = [] 50 | for i in range(0, h, scale): 51 | for j in range(0, w, scale): 52 | y = random.randint(-scale // 2, scale // 2) + i 53 | x = random.randint(-scale // 2, scale // 2) + j 54 | order.append((y % h, x % w)) 55 | return order 56 | '''prewitt算子''' 57 | def prewitt(self, img): 58 | img_gaussian = cv2.GaussianBlur(img, (3, 3), 0) 59 | kernelx = np.array([[1, 1, 1], [0, 0, 0], [-1, -1, -1]]) 60 | kernely = np.array([[-1, 0, 1], [-1, 0, 1], [-1, 0, 1]]) 61 | img_prewittx = cv2.filter2D(img_gaussian, -1, kernelx) 62 | img_prewitty = cv2.filter2D(img_gaussian, -1, kernely) 63 | return img_prewittx // 15.36, img_prewitty // 15.36 64 | '''roberts算子''' 65 | def roberts(self, img): 66 | roberts_cross_v = np.array([[0, 0, 0], [0, 1, 0], [0, 0, -1]]) 67 | roberts_cross_h = np.array([[0, 0, 0], [0, 0, 1], [0, -1, 0]]) 68 | vertical = ndimage.convolve(img, roberts_cross_v) 69 | horizontal = ndimage.convolve(img, roberts_cross_h) 70 | return vertical // 50.0, horizontal // 50.0 71 | '''利用边缘检测算子获得梯度''' 72 | def getgradient(self, img_o, ksize, edge_operator): 73 | if edge_operator == 'scharr': 74 | X = cv2.Scharr(img_o, cv2.CV_32F, 1, 0) / 50.0 75 | Y = cv2.Scharr(img_o, cv2.CV_32F, 0, 1) / 50.0 76 | elif edge_operator == 'prewitt': 77 | X, Y = self.prewitt(img_o) 78 | elif edge_operator == 'sobel': 79 | X = cv2.Sobel(img_o, cv2.CV_32F, 1, 0, ksize=5) / 50.0 80 | Y = cv2.Sobel(img_o, cv2.CV_32F, 0, 1, ksize=5) / 50.0 81 | elif edge_operator == 'roberts': 82 | X, Y = self.roberts(img_o) 83 | X = cv2.GaussianBlur(X, ksize, 0) 84 | Y = cv2.GaussianBlur(Y, ksize, 0) 85 | return X, Y -------------------------------------------------------------------------------- /pydrawing/modules/beautifiers/pencildrawing/__init__.py: -------------------------------------------------------------------------------- 1 | '''initialize''' 2 | from .pencildrawing import PencilDrawingBeautifier -------------------------------------------------------------------------------- /pydrawing/modules/beautifiers/pencildrawing/pencildrawing.py: -------------------------------------------------------------------------------- 1 | ''' 2 | Function: 3 | 复现论文"Combining Sketch and Tone for Pencil Drawing Production" 4 | Author: 5 | Charles 6 | 微信公众号: 7 | Charles的皮卡丘 8 | ''' 9 | import os 10 | import cv2 11 | import math 12 | import numpy as np 13 | from PIL import Image 14 | from scipy import signal 15 | from ..base import BaseBeautifier 16 | from scipy.ndimage import interpolation 17 | from scipy.sparse.linalg import spsolve 18 | from scipy.sparse import csr_matrix, spdiags 19 | 20 | 21 | '''图像处理工具''' 22 | class ImageProcessor(): 23 | '''将像素值压缩到[0, 1]''' 24 | @staticmethod 25 | def im2double(img): 26 | if len(img.shape) == 2: return (img - img.min()) / (img.max() - img.min()) 27 | else: return cv2.normalize(img.astype('float'), None, 0.0, 1.0, cv2.NORM_MINMAX) 28 | '''拉普拉斯分布''' 29 | @staticmethod 30 | def Laplace(x, sigma=9): 31 | value = (1. / sigma) * math.exp(-(256 - x) / sigma) * (256 - x) 32 | return value 33 | '''均匀分布''' 34 | @staticmethod 35 | def Uniform(x, ua=105, ub=225): 36 | value = (1. / (ub - ua)) * (max(x - ua, 0) - max(x - ub, 0)) 37 | return value 38 | '''高斯分布''' 39 | @staticmethod 40 | def Gaussian(x, u=90, sigma=11): 41 | value = (1. / math.sqrt(2 * math.pi * sigma)) * math.exp(-((x - u) ** 2) / (2 * (sigma ** 2))) 42 | return value 43 | '''水平方向拼接''' 44 | @staticmethod 45 | def horizontalStitch(img, width): 46 | img_stitch = img.copy() 47 | while img_stitch.shape[1] < width: 48 | window_size = int(round(img.shape[1] / 4.)) 49 | left = img[:, (img.shape[1]-window_size): img.shape[1]] 50 | right = img[:, :window_size] 51 | aleft = np.zeros((left.shape[0], window_size)) 52 | aright = np.zeros((left.shape[0], window_size)) 53 | for i in range(window_size): 54 | aleft[:, i] = left[:, i] * (1 - (i + 1.) / window_size) 55 | aright[:, i] = right[:, i] * (i + 1.) / window_size 56 | img_stitch = np.column_stack((img_stitch[:, :(img_stitch.shape[1]-window_size)], aleft+aright, img_stitch[:, window_size: img_stitch.shape[1]])) 57 | img_stitch = img_stitch[:, :width] 58 | return img_stitch 59 | '''垂直方向拼接''' 60 | @staticmethod 61 | def verticalStitch(img, height): 62 | img_stitch = img.copy() 63 | while img_stitch.shape[0] < height: 64 | window_size = int(round(img.shape[0] / 4.)) 65 | up = img[(img.shape[0]-window_size): img.shape[0], :] 66 | down = img[0:window_size, :] 67 | aup = np.zeros((window_size, up.shape[1])) 68 | adown = np.zeros((window_size, up.shape[1])) 69 | for i in range(window_size): 70 | aup[i, :] = up[i, :] * (1 - (i + 1.) / window_size) 71 | adown[i, :] = down[i, :] * (i + 1.) / window_size 72 | img_stitch = np.row_stack((img_stitch[:img_stitch.shape[0]-window_size, :], aup+adown, img_stitch[window_size: img_stitch.shape[0], :])) 73 | img_stitch = img_stitch[:height, :] 74 | return img_stitch 75 | 76 | 77 | '''复现论文"Combining Sketch and Tone for Pencil Drawing Production"''' 78 | class PencilDrawingBeautifier(BaseBeautifier): 79 | def __init__(self, mode='gray', kernel_size_scale=1/40, stroke_width=1, color_depth=1, weights_color=[62, 30, 5], weights_gray=[76, 22, 2], texture_path=None, **kwargs): 80 | super(PencilDrawingBeautifier, self).__init__(**kwargs) 81 | assert mode in ['gray', 'color'] 82 | self.rootdir = os.path.split(os.path.abspath(__file__))[0] 83 | self.image_processor = ImageProcessor() 84 | self.mode = mode 85 | # 铅笔笔画相关参数 86 | self.kernel_size_scale, self.stroke_width = kernel_size_scale, stroke_width 87 | # 铅笔色调相关参数 88 | self.weights_color, self.weights_gray, self.color_depth = weights_color, weights_gray, color_depth 89 | if (texture_path is None) or (not os.path.exists(texture_path)): self.texture_path = os.path.join(self.rootdir, 'textures/default.jpg') 90 | '''迭代图片''' 91 | def iterimage(self, image): 92 | if self.mode == 'color': 93 | img = Image.fromarray(image) 94 | img_ycbcr = img.convert('YCbCr') 95 | img = np.ndarray((img.size[1], img.size[0], 3), 'u1', img_ycbcr.tobytes()) 96 | img_out = img.copy() 97 | img_out.flags.writeable = True 98 | img_out[:, :, 0] = self.__strokeGeneration(img[:, :, 0]) * self.__toneGeneration(img[:, :, 0]) * 255 99 | img_out = cv2.cvtColor(img_out, cv2.COLOR_YCR_CB2BGR) 100 | else: 101 | img = image 102 | img = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY) 103 | img_s = self.__strokeGeneration(img) 104 | img_t = self.__toneGeneration(img) 105 | img_out = img_s * img_t * 255 106 | return img_out 107 | '''铅笔笔画生成''' 108 | def __strokeGeneration(self, img): 109 | h, w = img.shape 110 | kernel_size = int(min(w, h) * self.kernel_size_scale) 111 | kernel_size += kernel_size % 2 112 | # 计算梯度，产生幅度 113 | img_double = self.image_processor.im2double(img) 114 | dx = np.concatenate((np.abs(img_double[:, :-1]-img_double[:, 1:]), np.zeros((h, 1))), 1) 115 | dy = np.concatenate((np.abs(img_double[:-1, :]-img_double[1:, :]), np.zeros((1, w))), 0) 116 | img_gradient = np.sqrt(np.power(dx, 2) + np.power(dy, 2)) 117 | # 选择八个参考方向 118 | line_segments = np.zeros((kernel_size, kernel_size, 8)) 119 | for i in [0, 1, 2, 7]: 120 | for x in range(kernel_size): 121 | y = round((x + 1 - kernel_size / 2) * math.tan(math.pi / 8 * i)) 122 | y = kernel_size / 2 - y 123 | if y > 0 and y <= kernel_size: 124 | line_segments[int(y-1), x, i] = 1 125 | if i == 7: 126 | line_segments[:, :, 3] = np.rot90(line_segments[:, :, 7], -1) 127 | else: 128 | line_segments[:, :, i+4] = np.rot90(line_segments[:, :, i], 1) 129 | # 获取参考方向的响应图 130 | response_maps = np.zeros((h, w, 8)) 131 | for i in range(8): 132 | response_maps[:, :, i] = signal.convolve2d(img_gradient, line_segments[:, :, i], 'same') 133 | response_maps_maxvalueidx = response_maps.argmax(axis=-1) 134 | # 通过在所有方向的响应中选择最大值来进行分类 135 | magnitude_maps = np.zeros_like(response_maps) 136 | for i in range(8): 137 | magnitude_maps[:, :, i] = img_gradient * (response_maps_maxvalueidx == i).astype('float') 138 | # 线条整形 139 | stroke_maps = np.zeros_like(response_maps) 140 | for i in range(8): 141 | stroke_maps[:, :, i] = signal.convolve2d(magnitude_maps[:, :, i], line_segments[:, :, i], 'same') 142 | stroke_maps = stroke_maps.sum(axis=-1) 143 | stroke_maps = (stroke_maps - stroke_maps.min()) / (stroke_maps.max() - stroke_maps.min()) 144 | stroke_maps = (1 - stroke_maps) ** self.stroke_width 145 | return stroke_maps 146 | '''铅笔色调生成''' 147 | def __toneGeneration(self, img, mode=None): 148 | height, width = img.shape 149 | # 直方图匹配 150 | img_hist_match = self.__histogramMatching(img, mode) ** self.color_depth 151 | # 获得纹理 152 | texture = cv2.imread(self.texture_path) 153 | texture = cv2.cvtColor(texture, cv2.COLOR_BGR2GRAY)[99: texture.shape[0]-100, 99: texture.shape[1]-100] 154 | ratio = 0.2 * min(img.shape[0], img.shape[1]) / float(1024) 155 | texture = interpolation.zoom(texture, (ratio, ratio)) 156 | texture = self.image_processor.im2double(texture) 157 | texture = self.image_processor.horizontalStitch(texture, img.shape[1]) 158 | texture = self.image_processor.verticalStitch(texture, img.shape[0]) 159 | size = img.size 160 | nzmax = 2 * (size-1) 161 | i = np.zeros((nzmax, 1)) 162 | j = np.zeros((nzmax, 1)) 163 | s = np.zeros((nzmax, 1)) 164 | for m in range(1, nzmax+1): 165 | i[m-1] = int(math.ceil((m + 0.1) / 2)) - 1 166 | j[m-1] = int(math.ceil((m - 0.1) / 2)) - 1 167 | s[m-1] = -2 * (m % 2) + 1 168 | dx = csr_matrix((s.T[0], (i.T[0], j.T[0])), shape=(size, size)) 169 | nzmax = 2 * (size - img.shape[1]) 170 | i = np.zeros((nzmax, 1)) 171 | j = np.zeros((nzmax, 1)) 172 | s = np.zeros((nzmax, 1)) 173 | for m in range(1, nzmax+1): 174 | i[m-1, :] = int(math.ceil((m - 1 + 0.1) / 2) + img.shape[1] * (m % 2)) - 1 175 | j[m-1, :] = math.ceil((m - 0.1) / 2) - 1 176 | s[m-1, :] = -2 * (m % 2) + 1 177 | dy = csr_matrix((s.T[0], (i.T[0], j.T[0])), shape=(size, size)) 178 | texture_sparse = spdiags(np.log(np.reshape(texture.T, (1, texture.size), order="f") + 0.01), 0, size, size) 179 | img_hist_match1d = np.log(np.reshape(img_hist_match.T, (1, img_hist_match.size), order="f").T + 0.01) 180 | nat = texture_sparse.T.dot(img_hist_match1d) 181 | a = np.dot(texture_sparse.T, texture_sparse) 182 | b = dx.T.dot(dx) 183 | c = dy.T.dot(dy) 184 | mat = a + 0.2 * (b + c) 185 | beta1d = spsolve(mat, nat) 186 | beta = np.reshape(beta1d, (img.shape[0], img.shape[1]), order="c") 187 | tone = texture ** beta 188 | tone = (tone - tone.min()) / (tone.max() - tone.min()) 189 | return tone 190 | '''直方图匹配''' 191 | def __histogramMatching(self, img, mode=None): 192 | weights = self.weights_color if mode == 'color' else self.weights_gray 193 | # 图像 194 | histogram_img = cv2.calcHist([img], [0], None, [256], [0, 256]) 195 | histogram_img.resize(histogram_img.size) 196 | histogram_img /= histogram_img.sum() 197 | histogram_img_cdf = np.cumsum(histogram_img) 198 | # 自然图像 199 | histogram_natural = np.zeros_like(histogram_img) 200 | for x in range(256): 201 | histogram_natural[x] = weights[0] * self.image_processor.Laplace(x) + weights[1] * self.image_processor.Uniform(x) + weights[2] * self.image_processor.Gaussian(x) 202 | histogram_natural /= histogram_natural.sum() 203 | histogram_natural_cdf = np.cumsum(histogram_natural) 204 | # 做直方图匹配 205 | img_hist_match = np.zeros_like(img) 206 | for x in range(img.shape[0]): 207 | for y in range(img.shape[1]): 208 | value = histogram_img_cdf[img[x, y]] 209 | img_hist_match[x, y] = (np.abs(histogram_natural_cdf-value)).argmin() 210 | img_hist_match = np.true_divide(img_hist_match, 255) 211 | return img_hist_match -------------------------------------------------------------------------------- /pydrawing/modules/beautifiers/pencildrawing/textures/default.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/CharlesPikachu/pydrawing/e980ad9bf4cece42ff40ed2bc7bed7a155eabb8e/pydrawing/modules/beautifiers/pencildrawing/textures/default.jpg -------------------------------------------------------------------------------- /pydrawing/modules/beautifiers/photocorrection/__init__.py: -------------------------------------------------------------------------------- 1 | '''initialize''' 2 | from .photocorrection import PhotocorrectionBeautifier -------------------------------------------------------------------------------- /pydrawing/modules/beautifiers/photocorrection/photocorrection.py: -------------------------------------------------------------------------------- 1 | ''' 2 | Function: 3 | 简单的照片矫正 4 | Author: 5 | Charles 6 | 微信公众号: 7 | Charles的皮卡丘 8 | ''' 9 | import cv2 10 | import numpy as np 11 | from ..base import BaseBeautifier 12 | from imutils.perspective import four_point_transform 13 | 14 | 15 | '''简单的照片矫正''' 16 | class PhotocorrectionBeautifier(BaseBeautifier): 17 | def __init__(self, epsilon_factor=0.08, canny_boundaries=[100, 200], use_preprocess=False, **kwargs): 18 | super(PhotocorrectionBeautifier, self).__init__(**kwargs) 19 | self.epsilon_factor = epsilon_factor 20 | self.canny_boundaries = canny_boundaries 21 | self.use_preprocess = use_preprocess 22 | '''迭代图片''' 23 | def iterimage(self, image): 24 | # 预处理 25 | if self.use_preprocess: 26 | image_edge = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY) 27 | image_edge = cv2.GaussianBlur(image_edge, (5, 5), 0) 28 | image_edge = cv2.dilate(image_edge, cv2.getStructuringElement(cv2.MORPH_RECT, (3, 3))) 29 | else: 30 | image_edge = image.copy() 31 | image_edge = cv2.Canny(image_edge, self.canny_boundaries[0], self.canny_boundaries[1], 3) 32 | # 找到最大轮廓 33 | cnts = cv2.findContours(image_edge.copy(), cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE) 34 | cnts = cnts[0] 35 | if len(cnts) < 1: return image 36 | cnts = sorted(cnts, key=cv2.contourArea, reverse=True) 37 | for cnt in cnts: 38 | peri = cv2.arcLength(cnt, True) 39 | approx = cv2.approxPolyDP(cnt, self.epsilon_factor * peri, True) 40 | if len(approx) == 4: break 41 | if len(approx) != 4: return image 42 | # 矫正 43 | image_processed = four_point_transform(image, approx.reshape(4, 2)) 44 | # 返回 45 | return image_processed -------------------------------------------------------------------------------- /pydrawing/modules/beautifiers/photomosaic/__init__.py: -------------------------------------------------------------------------------- 1 | '''initialize''' 2 | from .photomosaic import PhotomosaicBeautifier -------------------------------------------------------------------------------- /pydrawing/modules/beautifiers/photomosaic/photomosaic.py: -------------------------------------------------------------------------------- 1 | ''' 2 | Function: 3 | 拼马赛克图片 4 | Author: 5 | Charles 6 | 微信公众号: 7 | Charles的皮卡丘 8 | ''' 9 | import cv2 10 | import glob 11 | import numpy as np 12 | from tqdm import tqdm 13 | from itertools import product 14 | from ..base import BaseBeautifier 15 | 16 | 17 | '''拼马赛克图片''' 18 | class PhotomosaicBeautifier(BaseBeautifier): 19 | def __init__(self, block_size=15, src_images_dir=None, **kwargs): 20 | super(PhotomosaicBeautifier, self).__init__(**kwargs) 21 | self.block_size = block_size 22 | self.src_images_dir = src_images_dir 23 | self.src_images, self.avg_colors = self.ReadSourceImages() 24 | '''迭代图片''' 25 | def iterimage(self, image): 26 | output_image = np.zeros(image.shape, np.uint8) 27 | src_images, avg_colors = self.src_images, self.avg_colors 28 | for i, j in tqdm(product(range(int(image.shape[1]/self.block_size)), range(int(image.shape[0]/self.block_size)))): 29 | block = image[j*self.block_size: (j+1)*self.block_size, i*self.block_size: (i+1)*self.block_size, :] 30 | avg_color = np.sum(np.sum(block, axis=0), axis=0) / (self.block_size * self.block_size) 31 | distances = np.linalg.norm(avg_color - avg_colors, axis=1) 32 | idx = np.argmin(distances) 33 | output_image[j*self.block_size: (j+1)*self.block_size, i*self.block_size: (i+1)*self.block_size, :] = src_images[idx] 34 | return output_image 35 | '''读取所有源图片并计算对应的颜色平均值''' 36 | def ReadSourceImages(self): 37 | src_images, avg_colors = [], [] 38 | for path in tqdm(glob.glob("{}/*[jpg,jpeg,png,gif]".format(self.src_images_dir))): 39 | image = cv2.imread(path, cv2.IMREAD_COLOR) 40 | if image.shape[-1] != 3: continue 41 | image = cv2.resize(image, (self.block_size, self.block_size)) 42 | avg_color = np.sum(np.sum(image, axis=0), axis=0) / (self.block_size * self.block_size) 43 | src_images.append(image) 44 | avg_colors.append(avg_color) 45 | return src_images, np.array(avg_colors) -------------------------------------------------------------------------------- /pydrawing/modules/utils/__init__.py: -------------------------------------------------------------------------------- 1 | '''initialize''' 2 | from .logger import Logger 3 | from .io import Images2VideoAndSave, SaveImage, ReadVideo, SaveImage, checkdir -------------------------------------------------------------------------------- /pydrawing/modules/utils/io.py: -------------------------------------------------------------------------------- 1 | ''' 2 | Function: 3 | IO相关的工具函数 4 | Author: 5 | Charles 6 | 微信公众号: 7 | Charles的皮卡丘 8 | ''' 9 | import os 10 | import cv2 11 | from tqdm import tqdm 12 | 13 | 14 | '''检查文件是否存在''' 15 | def checkdir(dirname): 16 | if os.path.exists(dirname): return True 17 | os.mkdir(dirname) 18 | return False 19 | 20 | 21 | '''将图片转为视频并保存''' 22 | def Images2VideoAndSave(images, savedir='outputs', savename='output', fps=25, ext='avi', logger_handle=None): 23 | checkdir(savedir) 24 | savepath = os.path.join(savedir, savename + f'.{ext}') 25 | fourcc = cv2.VideoWriter_fourcc('M', 'J', 'P', 'G') 26 | video_writer = cv2.VideoWriter(savepath, fourcc, fps, (images[0].shape[1], images[0].shape[0])) 27 | pbar = tqdm(images) 28 | for image in pbar: 29 | pbar.set_description(f'Writing image to {savepath}') 30 | video_writer.write(image) 31 | if logger_handle is not None: logger_handle.info(f'Video is saved into {savepath}') 32 | 33 | 34 | '''保存图片''' 35 | def SaveImage(image, savedir='outputs', savename='output', ext='png', logger_handle=None): 36 | checkdir(savedir) 37 | savepath = os.path.join(savedir, savename + f'.{ext}') 38 | cv2.imwrite(savepath, image) 39 | if logger_handle is not None: logger_handle.info(f'Image is saved into {savepath}') 40 | 41 | 42 | '''读取视频''' 43 | def ReadVideo(videopath): 44 | capture, images = cv2.VideoCapture(videopath), [] 45 | fps = capture.get(cv2.CAP_PROP_FPS) 46 | while capture.isOpened(): 47 | ret, frame = capture.read() 48 | if not ret: break 49 | images.append(frame) 50 | capture.release() 51 | return images, fps -------------------------------------------------------------------------------- /pydrawing/modules/utils/logger.py: -------------------------------------------------------------------------------- 1 | ''' 2 | Function: 3 | 一些终端打印工具 4 | Author: 5 | Charles 6 | 微信公众号: 7 | Charles的皮卡丘 8 | ''' 9 | import logging 10 | 11 | 12 | '''打印日志类''' 13 | class Logger(): 14 | def __init__(self, logfilepath, **kwargs): 15 | setattr(self, 'logfilepath', logfilepath) 16 | logging.basicConfig( 17 | level=logging.INFO, 18 | format='%(asctime)s %(levelname)-8s %(message)s', 19 | datefmt='%Y-%m-%d %H:%M:%S', 20 | handlers=[logging.FileHandler(logfilepath), logging.StreamHandler()], 21 | ) 22 | @staticmethod 23 | def log(level, message): 24 | logging.log(level, message) 25 | def debug(self, message, disable_print=False): 26 | if disable_print: 27 | fp = open(self.logfilepath, 'a') 28 | fp.write(message + '\n') 29 | else: 30 | Logger.log(logging.DEBUG, message) 31 | def info(self, message, disable_print=False): 32 | if disable_print: 33 | fp = open(self.logfilepath, 'a') 34 | fp.write(message + '\n') 35 | else: 36 | Logger.log(logging.INFO, message) 37 | def warning(self, message, disable_print=False): 38 | if disable_print: 39 | fp = open(self.logfilepath, 'a') 40 | fp.write(message + '\n') 41 | else: 42 | Logger.log(logging.WARNING, message) 43 | def error(self, message, disable_print=False): 44 | if disable_print: 45 | fp = open(self.logfilepath, 'a') 46 | fp.write(message + '\n') 47 | else: 48 | Logger.log(logging.ERROR, message) -------------------------------------------------------------------------------- /pydrawing/pydrawing.py: -------------------------------------------------------------------------------- 1 | ''' 2 | Function: 3 | 用Python美化你的照片或视频 4 | Author: 5 | Charles 6 | 微信公众号: 7 | Charles的皮卡丘 8 | ''' 9 | import warnings 10 | if __name__ == '__main__': 11 | from modules import * 12 | else: 13 | from .modules import * 14 | warnings.filterwarnings('ignore') 15 | 16 | 17 | '''用Python美化你的照片或视频''' 18 | class pydrawing(): 19 | def __init__(self, **kwargs): 20 | for key, value in kwargs.items(): setattr(self, key, value) 21 | self.supported_beautifiers = self.initializebeautifiers() 22 | self.logger_handle = Logger(kwargs.get('logfilepath', 'pydrawing.log')) 23 | print(self) 24 | '''执行对应的算法''' 25 | def execute(self, filepath='asserts/dog.jpg', beautifier_type=None, config={}): 26 | assert beautifier_type in self.supported_beautifiers, 'unsupport beautifier_type %s' % beautifier_type 27 | if 'savedir' not in config: config['savedir'] = 'outputs' 28 | if 'savename' not in config: config['savename'] = 'output' 29 | if 'logger_handle' not in config: config['logger_handle'] = self.logger_handle 30 | beautifier = self.supported_beautifiers[beautifier_type](**config) 31 | beautifier.process(filepath) 32 | '''获得所有支持的美化器''' 33 | def getallsupports(self): 34 | return list(self.supported_beautifiers.keys()) 35 | '''初始化美化器''' 36 | def initializebeautifiers(self): 37 | supported_beautifiers = { 38 | 'glitch': GlitchBeautifier, 39 | 'cartoonise': CartooniseBeautifier, 40 | 'cartoongan': CartoonGanBeautifier, 41 | 'oilpainting': OilpaintingBeautifier, 42 | 'beziercurve': BezierCurveBeautifier, 43 | 'photomosaic': PhotomosaicBeautifier, 44 | 'characterize': CharacterizeBeautifier, 45 | 'douyineffect': DouyinEffectBeautifier, 46 | 'noteprocessor': NoteprocessorBeautifier, 47 | 'pencildrawing': PencilDrawingBeautifier, 48 | 'cartoonizeface': CartoonizeFaceBeautifier, 49 | 'nostalgicstyle': NostalgicstyleBeautifier, 50 | 'photocorrection': PhotocorrectionBeautifier, 51 | 'geneticfittingcircle': GeneticFittingCircleBeautifier, 52 | 'geneticfittingpolygon': GeneticFittingPolygonBeautifier, 53 | 'fastneuralstyletransfer': FastNeuralStyleTransferBeautifier, 54 | } 55 | return supported_beautifiers 56 | '''str''' 57 | def __str__(self): 58 | return 'Welcome to use Pydrawing!\nYou can visit https://github.com/CharlesPikachu/pydrawing for more details.' 59 | 60 | 61 | '''run''' 62 | if __name__ == '__main__': 63 | import random 64 | drawing_client = pydrawing() 65 | drawing_client.execute('asserts/dog.jpg', random.choice(drawing_client.getallsupports())) -------------------------------------------------------------------------------- /requirements.txt: -------------------------------------------------------------------------------- 1 | opencv-python 2 | numpy 3 | tqdm 4 | pillow 5 | beautifulsoup4 6 | imutils -------------------------------------------------------------------------------- /setup.py: -------------------------------------------------------------------------------- 1 | ''' 2 | Function: 3 | setup the pydrawing 4 | Author: 5 | Charles 6 | 微信公众号: 7 | Charles的皮卡丘 8 | GitHub: 9 | https://github.com/CharlesPikachu 10 | ''' 11 | import pydrawing 12 | from setuptools import setup, find_packages 13 | 14 | 15 | '''readme''' 16 | with open('README.md', 'r', encoding='utf-8') as f: 17 | long_description = f.read() 18 | 19 | 20 | '''package data''' 21 | package_data = {} 22 | package_data.update({ 23 | 'pydrawing.modules.beautifiers.pencildrawing': ['textures/*'] 24 | }) 25 | package_data.update({ 26 | 'pydrawing.modules.beautifiers.beziercurve': ['potrace.exe'] 27 | }) 28 | 29 | 30 | '''setup''' 31 | setup( 32 | name=pydrawing.__title__, 33 | version=pydrawing.__version__, 34 | description=pydrawing.__description__, 35 | long_description=long_description, 36 | long_description_content_type='text/markdown', 37 | classifiers=[ 38 | 'License :: OSI Approved :: Apache Software License', 39 | 'Programming Language :: Python :: 3', 40 | 'Intended Audience :: Developers', 41 | 'Operating System :: OS Independent' 42 | ], 43 | author=pydrawing.__author__, 44 | url=pydrawing.__url__, 45 | author_email=pydrawing.__email__, 46 | license=pydrawing.__license__, 47 | include_package_data=True, 48 | package_data=package_data, 49 | install_requires=[lab.strip('\n') for lab in list(open('requirements.txt', 'r').readlines())], 50 | zip_safe=True, 51 | packages=find_packages(), 52 | ) --------------------------------------------------------------------------------