├── test ├── WechatIMG383.jpg ├── WechatIMG389.jpg └── WechatIMG390.jpg ├── label_image.py ├── README.md ├── LICENSE └── retrain.py /test/WechatIMG383.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Anymake/tensorflow_flow_demo/HEAD/test/WechatIMG383.jpg -------------------------------------------------------------------------------- /test/WechatIMG389.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Anymake/tensorflow_flow_demo/HEAD/test/WechatIMG389.jpg -------------------------------------------------------------------------------- /test/WechatIMG390.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Anymake/tensorflow_flow_demo/HEAD/test/WechatIMG390.jpg -------------------------------------------------------------------------------- /label_image.py: -------------------------------------------------------------------------------- 1 | import os, sys 2 | 3 | import tensorflow as tf 4 | 5 | os.environ['TF_CPP_MIN_LOG_LEVEL'] = '2' 6 | 7 | # change this as you see fit 8 | image_path = sys.argv[1] 9 | 10 | # Read in the image_data 11 | image_data = tf.gfile.FastGFile(image_path, 'rb').read() 12 | 13 | # Loads label file, strips off carriage return 14 | label_lines = [line.rstrip() for line 15 | in tf.gfile.GFile("retrained_labels.txt")] 16 | 17 | # Unpersists graph from file 18 | with tf.gfile.FastGFile("retrained_graph.pb", 'rb') as f: 19 | graph_def = tf.GraphDef() 20 | graph_def.ParseFromString(f.read()) 21 | tf.import_graph_def(graph_def, name='') 22 | 23 | with tf.Session() as sess: 24 | # Feed the image_data as input to the graph and get first prediction 25 | softmax_tensor = sess.graph.get_tensor_by_name('final_result:0') 26 | 27 | predictions = sess.run(softmax_tensor, \ 28 | {'DecodeJpeg/contents:0': image_data}) 29 | 30 | # Sort to show labels of first prediction in order of confidence 31 | top_k = predictions[0].argsort()[-len(predictions[0]):][::-1] 32 | 33 | for node_id in top_k: 34 | human_string = label_lines[node_id] 35 | score = predictions[0][node_id] 36 | print('%s (score = %.5f)' % (human_string, score)) -------------------------------------------------------------------------------- /README.md: -------------------------------------------------------------------------------- 1 | 2 | # 基于TensorFlow训练花朵识别模型的源码和Demo 3 | 下面就通过对现有的 Google Inception-V3 模型进行 retrain ，对 5 种花朵样本数据的进行训练，来完成一个可以识别五种花朵的模型，并将新训练的模型进行测试部属，让大家体验一下完整的流程。 4 | 5 | ![花朵训练样本](https://img-blog.csdn.net/20180602195623764?watermark/2/text/aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L0FueW1ha2VfcmVu/font/5a6L5L2T/fontsize/400/fill/I0JBQkFCMA==/dissolve/70) 6 | 7 | 8 | ### 安装 TensorFlow （Mac 为例） 9 | 10 | 其他平台可以直接参考官网说明：[Installing TensorFlow](https://www.tensorflow.org/install/) 11 | 12 | #### 首先检查系统是否安装了 Python 13 | 14 | 要安装 `TensorFlow` ，你的系统必须依据安装了以下任一 `Python` 版本： 15 | 16 | * **Python 2.7** 17 | * **Python 3.3+** 18 | 19 | 如果做数据处理较多的话，建议安装Anaconda， **Anaconda** 是一种Python语言的免费增值开源发行版 ，用于进行大规模数据处理, 预测分析, 和科学计算, 致力于简化包的管理和部署。Anaconda使用软件包管理系统Conda进行包管理。安装完成后输入shell下输入`python`即可查看Anaconda对应的Python 版本，我使用的是Python 2.7.14： 20 | ``` 21 | ➜ ~ python 22 | Python 2.7.14 |Anaconda, Inc.| (default, Dec 7 2017, 11:07:58) 23 | [GCC 4.2.1 Compatible Clang 4.0.1 (tags/RELEASE_401/final)] on darwin 24 | Type "help", "copyright", "credits" or "license" for more information. 25 | 26 | ``` 27 | 如果你的系统还没有安装符合以上版本的 Python，现在安装。 28 | 29 | 30 | 31 | #### 通过 pip 安装 TensorFlow 32 | 33 | ``` 34 | # Python 2 35 | ➜ pip install tensorflow 36 | # Python 3 37 | ➜ pip3 install tensorflow 38 | 39 | ``` 40 | 41 | #### 通过官方样例测试 TensorFlow 是否正常安装 42 | 43 | 进入 Python 环境后输入以下代码，当出现 `“Hello, TensorFlow!”` 时表明已经安装成功，可正常使用 TensorFlow 了。 44 | 45 | ``` 46 | ➜ python 47 | import tensorflow as tf 48 | hello = tf.constant('Hello, TensorFlow!') 49 | sess = tf.Session() 50 | print(sess.run(hello)) 51 | Hello, TensorFlow! 52 | 53 | ``` 54 | 55 | ### 准备训练样本 56 | 57 | 现在我们要训练花朵的识别模型，这是 Google 在TensorFlow里面提供的一个例子，其中包含了5类花朵的训练图片。可以新建个flower_demo文件夹，用于存放数据和训练的模型。 58 | 59 | **下载并解压得到训练样本** 60 | 61 | ``` 62 | cd flower_demo 63 | # 下载和解压花朵训练数据 64 | curl -O http://download.tensorflow.org/example_images/flower_photos.tgz 65 | tar xzf flower_photos.tgz 66 | 67 | ``` 68 | 69 | 打开训练样本文件夹 flower_photos ，里面有 5 种类别的花：`daisy(雏菊), dandelion(蒲公英), roses(玫瑰), sunflowers(向日葵) , tulips(郁金香)`，总共3672张，每个类别的大概有 600-900 张训练样本图片,具体如下： 70 | 71 | ``` 72 | cd flower_photos 73 | for dir in `find ./ -maxdepth 1 -type d`;do echo -n -e "$dir\t";find $dir -type f|wc -l ;done; 74 | ./ 3672 75 | .//roses 641 76 | .//sunflowers 699 77 | .//daisy 633 78 | .//dandelion 898 79 | .//tulips 799 80 | 81 | ``` 82 | ### 开始训练 83 | 84 | **下载训练模型使用的 retrain 脚本** 85 | 该脚本会自动下载 google Inception v3 模型相关文件，`retrain.py` 是 Google 提供的以ImageNet图片分类模型为基础模型，利用flower_photos数据迁移训练花朵识别模型的脚本。 86 | 87 | ``` 88 | cd flower_demo 89 | curl -O https://raw.githubusercontent.com/tensorflow/tensorflow/r1.1/tensorflow/examples/image_retraining/retrain.py 90 | 91 | ``` 92 | **启动训练脚本，开始训练模型** 93 | 94 | 在运行 `retrain.py` 脚本时，需要配置一些运行命令参数，比如指定模型输入输出相关名称和其他训练要求的配置。其中`--how_many_training_steps=4000`配置代表训练迭代次数，默认值为4000，如果机器较差，可以适当减少这个值。 95 | 96 | ``` 97 | ➜ cd flower_demo 98 | ➜ python3 retrain.py \ 99 | --bottleneck_dir=bottlenecks \ 100 | --how_many_training_steps=4000 \ 101 | --model_dir=inception \ 102 | --summaries_dir=training_summaries/basic \ 103 | --output_graph=retrained_graph.pb \ 104 | --output_labels=retrained_labels.txt \ 105 | --image_dir=flower_photos 106 | 107 | ``` 108 | 这里我们训练4000steps，时间不是很久，我在配备4.2 GHz Intel Core i7处理器的iMac上，不适用GPU大概就5分钟就能训练完成。模型训练完成后，可以看到测试集上`Final test accuracy = 92.1%`，也就是说我们训练的5类花朵识别模型，在测试集上已经有92%的识别准确率了。其中生成的 `retrained_labels.txt` 和 `retrained_graph.pb` 这两个是模型相关文件。 109 | ``` 110 | 2018-06-02 15:47:00.266119: Step 3950: Train accuracy = 94.0% 111 | 2018-06-02 15:47:00.266159: Step 3950: Cross entropy = 0.135385 112 | 2018-06-02 15:47:00.327843: Step 3950: Validation accuracy = 93.0% (N=100) 113 | 2018-06-02 15:47:00.976543: Step 3960: Train accuracy = 94.0% 114 | 2018-06-02 15:47:00.976591: Step 3960: Cross entropy = 0.234760 115 | 2018-06-02 15:47:01.038559: Step 3960: Validation accuracy = 91.0% (N=100) 116 | 2018-06-02 15:47:01.667255: Step 3970: Train accuracy = 97.0% 117 | 2018-06-02 15:47:01.667372: Step 3970: Cross entropy = 0.167394 118 | 2018-06-02 15:47:01.731935: Step 3970: Validation accuracy = 87.0% (N=100) 119 | 2018-06-02 15:47:02.355780: Step 3980: Train accuracy = 96.0% 120 | 2018-06-02 15:47:02.355818: Step 3980: Cross entropy = 0.151201 121 | 2018-06-02 15:47:02.418314: Step 3980: Validation accuracy = 91.0% (N=100) 122 | 2018-06-02 15:47:03.042364: Step 3990: Train accuracy = 99.0% 123 | 2018-06-02 15:47:03.042402: Step 3990: Cross entropy = 0.094383 124 | 2018-06-02 15:47:03.103718: Step 3990: Validation accuracy = 91.0% (N=100) 125 | 2018-06-02 15:47:03.667861: Step 3999: Train accuracy = 99.0% 126 | 2018-06-02 15:47:03.667899: Step 3999: Cross entropy = 0.106797 127 | 2018-06-02 15:47:03.729215: Step 3999: Validation accuracy = 94.0% (N=100) 128 | Final test accuracy = 92.1% (N=353) 129 | ``` 130 | ### 测试训练完成后的模型 131 | 132 | 同样的，我们先下载测试模型的脚本 `label_image.py`，然后从flower_photos/daisy/文件夹下选择图片488202750_c420cbce61.jpg，测试我们训练后的模型的识别准确率，当然你也可以百度搜索一张5类花朵的任意一张图测试识别效果，从下图可以看出，我们训练的算法模型认为这张图属于`daisy`的概率高达98.9%. 133 | 134 | ``` 135 | ➜ cd flower_demo 136 | ➜ curl -L https://goo.gl/3lTKZs > label_image.py 137 | ➜ python label_image.py flower_photos/daisy/488202750_c420cbce61.jpg 138 | 139 | daisy (score = 0.98921) 140 | sunflowers (score = 0.00948) 141 | dandelion (score = 0.00088) 142 | tulips (score = 0.00038) 143 | roses (score = 0.00005) 144 | ``` 145 | ![蒲公英测试图](https://img-blog.csdn.net/20180602200253465?watermark/2/text/aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L0FueW1ha2VfcmVu/font/5a6L5L2T/fontsize/400/fill/I0JBQkFCMA==/dissolve/70) 146 | 有人说`label_image.py`无法下载，代码如下： 147 | ``` 148 | import os, sys 149 | import tensorflow as tf 150 | os.environ['TF_CPP_MIN_LOG_LEVEL'] = '2' 151 | 152 | # change this as you see fit 153 | image_path = sys.argv[1] 154 | 155 | # Read in the image_data 156 | image_data = tf.gfile.FastGFile(image_path, 'rb').read() 157 | 158 | # Loads label file, strips off carriage return 159 | label_lines = [line.rstrip() for line in tf.gfile.GFile("retrained_labels.txt")] 160 | 161 | # Unpersists graph from file 162 | with tf.gfile.FastGFile("retrained_graph.pb", 'rb') as f: 163 | graph_def = tf.GraphDef() 164 | graph_def.ParseFromString(f.read()) 165 | tf.import_graph_def(graph_def, name='') 166 | 167 | with tf.Session() as sess: 168 | # Feed the image_data as input to the graph and get first prediction 169 | softmax_tensor = sess.graph.get_tensor_by_name('final_result:0') 170 | 171 | predictions = sess.run(softmax_tensor, {'DecodeJpeg/contents:0': image_data}) 172 | 173 | # Sort to show labels of first prediction in order of confidence 174 | top_k = predictions[0].argsort()[-len(predictions[0]):][::-1] 175 | 176 | for node_id in top_k: 177 | human_string = label_lines[node_id] 178 | score = predictions[0][node_id] 179 | print('%s (score = %.5f)' % (human_string, score)) 180 | ``` 181 | 我们随便从百度搜索一张蒲公英（dandelion）的图，保存到`test/WechatIMG383.jpg`，测试结果显示属于蒲公英的概率为99.59%. 182 | 183 | ``` 184 | python label_image.py test/WechatIMG383.jpg 185 | 186 | dandelion (score = 0.99592) 187 | sunflowers (score = 0.00359) 188 | daisy (score = 0.00042) 189 | tulips (score = 0.00005) 190 | roses (score = 0.00001) 191 | ``` 192 | 以上基本是模型训练和测试的全部过程，希望能让大家对深度学习的完整项目有个大致的了解。 193 | 194 | **启动 TensorBoard** 195 | TensorBoard 是 TensorFlow 自带的训练效果可视化的分析工具，我们可以利用此工具检测和分析模型的收敛情况，比如查看loss的下降、acc的提升和查看可视化的网络结构图等。在我们建的工程目录下，启动tensorboard的具体命令如下： 196 | 197 | ``` 198 | ➜ cd flower_demo 199 | ➜ tensorboard --logdir training_summaries 200 | 201 | ``` 202 | 203 | 启动 TensorBoard 会占用系统 `6006` 端口 ，再启动一个新的 TensorBoard 之前，必须要 kill 已在运行的 TensorBoard 任务。 204 | 205 | ``` 206 | ➜ pkill -f "tensorboard 207 | 208 | ``` 209 | **启动浏览器查看 TensorBoard** 210 | 211 | 启动TensorBoard后，可以启动浏览器，在地址栏中输入 `localhost:6006` 来查看训练进度以及loss和准确度的变化，分析模型等。 212 | 213 | ![训练过程中loss和准确率的变化](https://img-blog.csdn.net/20180602200349392?watermark/2/text/aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L0FueW1ha2VfcmVu/font/5a6L5L2T/fontsize/400/fill/I0JBQkFCMA==/dissolve/70) 214 | 215 | 216 | ![花朵识别网络模型的后半部分](https://img-blog.csdn.net/20180602200413197?watermark/2/text/aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L0FueW1ha2VfcmVu/font/5a6L5L2T/fontsize/400/fill/I0JBQkFCMA==/dissolve/70) 217 | -------------------------------------------------------------------------------- /LICENSE: -------------------------------------------------------------------------------- 1 | GNU GENERAL PUBLIC LICENSE 2 | Version 3, 29 June 2007 3 | 4 | Copyright (C) 2007 Free Software Foundation, Inc. 5 | Everyone is permitted to copy and distribute verbatim copies 6 | of this license document, but changing it is not allowed. 7 | 8 | Preamble 9 | 10 | The GNU General Public License is a free, copyleft license for 11 | software and other kinds of works. 12 | 13 | The licenses for most software and other practical works are designed 14 | to take away your freedom to share and change the works. By contrast, 15 | the GNU General Public License is intended to guarantee your freedom to 16 | share and change all versions of a program--to make sure it remains free 17 | software for all its users. We, the Free Software Foundation, use the 18 | GNU General Public License for most of our software; it applies also to 19 | any other work released this way by its authors. You can apply it to 20 | your programs, too. 21 | 22 | When we speak of free software, we are referring to freedom, not 23 | price. Our General Public Licenses are designed to make sure that you 24 | have the freedom to distribute copies of free software (and charge for 25 | them if you wish), that you receive source code or can get it if you 26 | want it, that you can change the software or use pieces of it in new 27 | free programs, and that you know you can do these things. 28 | 29 | To protect your rights, we need to prevent others from denying you 30 | these rights or asking you to surrender the rights. Therefore, you have 31 | certain responsibilities if you distribute copies of the software, or if 32 | you modify it: responsibilities to respect the freedom of others. 33 | 34 | For example, if you distribute copies of such a program, whether 35 | gratis or for a fee, you must pass on to the recipients the same 36 | freedoms that you received. You must make sure that they, too, receive 37 | or can get the source code. And you must show them these terms so they 38 | know their rights. 39 | 40 | Developers that use the GNU GPL protect your rights with two steps: 41 | (1) assert copyright on the software, and (2) offer you this License 42 | giving you legal permission to copy, distribute and/or modify it. 43 | 44 | For the developers' and authors' protection, the GPL clearly explains 45 | that there is no warranty for this free software. For both users' and 46 | authors' sake, the GPL requires that modified versions be marked as 47 | changed, so that their problems will not be attributed erroneously to 48 | authors of previous versions. 49 | 50 | Some devices are designed to deny users access to install or run 51 | modified versions of the software inside them, although the manufacturer 52 | can do so. This is fundamentally incompatible with the aim of 53 | protecting users' freedom to change the software. The systematic 54 | pattern of such abuse occurs in the area of products for individuals to 55 | use, which is precisely where it is most unacceptable. Therefore, we 56 | have designed this version of the GPL to prohibit the practice for those 57 | products. If such problems arise substantially in other domains, we 58 | stand ready to extend this provision to those domains in future versions 59 | of the GPL, as needed to protect the freedom of users. 60 | 61 | Finally, every program is threatened constantly by software patents. 62 | States should not allow patents to restrict development and use of 63 | software on general-purpose computers, but in those that do, we wish to 64 | avoid the special danger that patents applied to a free program could 65 | make it effectively proprietary. To prevent this, the GPL assures that 66 | patents cannot be used to render the program non-free. 67 | 68 | The precise terms and conditions for copying, distribution and 69 | modification follow. 70 | 71 | TERMS AND CONDITIONS 72 | 73 | 0. Definitions. 74 | 75 | "This License" refers to version 3 of the GNU General Public License. 76 | 77 | "Copyright" also means copyright-like laws that apply to other kinds of 78 | works, such as semiconductor masks. 79 | 80 | "The Program" refers to any copyrightable work licensed under this 81 | License. Each licensee is addressed as "you". "Licensees" and 82 | "recipients" may be individuals or organizations. 83 | 84 | To "modify" a work means to copy from or adapt all or part of the work 85 | in a fashion requiring copyright permission, other than the making of an 86 | exact copy. The resulting work is called a "modified version" of the 87 | earlier work or a work "based on" the earlier work. 88 | 89 | A "covered work" means either the unmodified Program or a work based 90 | on the Program. 91 | 92 | To "propagate" a work means to do anything with it that, without 93 | permission, would make you directly or secondarily liable for 94 | infringement under applicable copyright law, except executing it on a 95 | computer or modifying a private copy. Propagation includes copying, 96 | distribution (with or without modification), making available to the 97 | public, and in some countries other activities as well. 98 | 99 | To "convey" a work means any kind of propagation that enables other 100 | parties to make or receive copies. Mere interaction with a user through 101 | a computer network, with no transfer of a copy, is not conveying. 102 | 103 | An interactive user interface displays "Appropriate Legal Notices" 104 | to the extent that it includes a convenient and prominently visible 105 | feature that (1) displays an appropriate copyright notice, and (2) 106 | tells the user that there is no warranty for the work (except to the 107 | extent that warranties are provided), that licensees may convey the 108 | work under this License, and how to view a copy of this License. If 109 | the interface presents a list of user commands or options, such as a 110 | menu, a prominent item in the list meets this criterion. 111 | 112 | 1. Source Code. 113 | 114 | The "source code" for a work means the preferred form of the work 115 | for making modifications to it. "Object code" means any non-source 116 | form of a work. 117 | 118 | A "Standard Interface" means an interface that either is an official 119 | standard defined by a recognized standards body, or, in the case of 120 | interfaces specified for a particular programming language, one that 121 | is widely used among developers working in that language. 122 | 123 | The "System Libraries" of an executable work include anything, other 124 | than the work as a whole, that (a) is included in the normal form of 125 | packaging a Major Component, but which is not part of that Major 126 | Component, and (b) serves only to enable use of the work with that 127 | Major Component, or to implement a Standard Interface for which an 128 | implementation is available to the public in source code form. A 129 | "Major Component", in this context, means a major essential component 130 | (kernel, window system, and so on) of the specific operating system 131 | (if any) on which the executable work runs, or a compiler used to 132 | produce the work, or an object code interpreter used to run it. 133 | 134 | The "Corresponding Source" for a work in object code form means all 135 | the source code needed to generate, install, and (for an executable 136 | work) run the object code and to modify the work, including scripts to 137 | control those activities. However, it does not include the work's 138 | System Libraries, or general-purpose tools or generally available free 139 | programs which are used unmodified in performing those activities but 140 | which are not part of the work. For example, Corresponding Source 141 | includes interface definition files associated with source files for 142 | the work, and the source code for shared libraries and dynamically 143 | linked subprograms that the work is specifically designed to require, 144 | such as by intimate data communication or control flow between those 145 | subprograms and other parts of the work. 146 | 147 | The Corresponding Source need not include anything that users 148 | can regenerate automatically from other parts of the Corresponding 149 | Source. 150 | 151 | The Corresponding Source for a work in source code form is that 152 | same work. 153 | 154 | 2. Basic Permissions. 155 | 156 | All rights granted under this License are granted for the term of 157 | copyright on the Program, and are irrevocable provided the stated 158 | conditions are met. This License explicitly affirms your unlimited 159 | permission to run the unmodified Program. The output from running a 160 | covered work is covered by this License only if the output, given its 161 | content, constitutes a covered work. This License acknowledges your 162 | rights of fair use or other equivalent, as provided by copyright law. 163 | 164 | You may make, run and propagate covered works that you do not 165 | convey, without conditions so long as your license otherwise remains 166 | in force. You may convey covered works to others for the sole purpose 167 | of having them make modifications exclusively for you, or provide you 168 | with facilities for running those works, provided that you comply with 169 | the terms of this License in conveying all material for which you do 170 | not control copyright. Those thus making or running the covered works 171 | for you must do so exclusively on your behalf, under your direction 172 | and control, on terms that prohibit them from making any copies of 173 | your copyrighted material outside their relationship with you. 174 | 175 | Conveying under any other circumstances is permitted solely under 176 | the conditions stated below. Sublicensing is not allowed; section 10 177 | makes it unnecessary. 178 | 179 | 3. Protecting Users' Legal Rights From Anti-Circumvention Law. 180 | 181 | No covered work shall be deemed part of an effective technological 182 | measure under any applicable law fulfilling obligations under article 183 | 11 of the WIPO copyright treaty adopted on 20 December 1996, or 184 | similar laws prohibiting or restricting circumvention of such 185 | measures. 186 | 187 | When you convey a covered work, you waive any legal power to forbid 188 | circumvention of technological measures to the extent such circumvention 189 | is effected by exercising rights under this License with respect to 190 | the covered work, and you disclaim any intention to limit operation or 191 | modification of the work as a means of enforcing, against the work's 192 | users, your or third parties' legal rights to forbid circumvention of 193 | technological measures. 194 | 195 | 4. Conveying Verbatim Copies. 196 | 197 | You may convey verbatim copies of the Program's source code as you 198 | receive it, in any medium, provided that you conspicuously and 199 | appropriately publish on each copy an appropriate copyright notice; 200 | keep intact all notices stating that this License and any 201 | non-permissive terms added in accord with section 7 apply to the code; 202 | keep intact all notices of the absence of any warranty; and give all 203 | recipients a copy of this License along with the Program. 204 | 205 | You may charge any price or no price for each copy that you convey, 206 | and you may offer support or warranty protection for a fee. 207 | 208 | 5. Conveying Modified Source Versions. 209 | 210 | You may convey a work based on the Program, or the modifications to 211 | produce it from the Program, in the form of source code under the 212 | terms of section 4, provided that you also meet all of these conditions: 213 | 214 | a) The work must carry prominent notices stating that you modified 215 | it, and giving a relevant date. 216 | 217 | b) The work must carry prominent notices stating that it is 218 | released under this License and any conditions added under section 219 | 7. This requirement modifies the requirement in section 4 to 220 | "keep intact all notices". 221 | 222 | c) You must license the entire work, as a whole, under this 223 | License to anyone who comes into possession of a copy. This 224 | License will therefore apply, along with any applicable section 7 225 | additional terms, to the whole of the work, and all its parts, 226 | regardless of how they are packaged. This License gives no 227 | permission to license the work in any other way, but it does not 228 | invalidate such permission if you have separately received it. 229 | 230 | d) If the work has interactive user interfaces, each must display 231 | Appropriate Legal Notices; however, if the Program has interactive 232 | interfaces that do not display Appropriate Legal Notices, your 233 | work need not make them do so. 234 | 235 | A compilation of a covered work with other separate and independent 236 | works, which are not by their nature extensions of the covered work, 237 | and which are not combined with it such as to form a larger program, 238 | in or on a volume of a storage or distribution medium, is called an 239 | "aggregate" if the compilation and its resulting copyright are not 240 | used to limit the access or legal rights of the compilation's users 241 | beyond what the individual works permit. Inclusion of a covered work 242 | in an aggregate does not cause this License to apply to the other 243 | parts of the aggregate. 244 | 245 | 6. Conveying Non-Source Forms. 246 | 247 | You may convey a covered work in object code form under the terms 248 | of sections 4 and 5, provided that you also convey the 249 | machine-readable Corresponding Source under the terms of this License, 250 | in one of these ways: 251 | 252 | a) Convey the object code in, or embodied in, a physical product 253 | (including a physical distribution medium), accompanied by the 254 | Corresponding Source fixed on a durable physical medium 255 | customarily used for software interchange. 256 | 257 | b) Convey the object code in, or embodied in, a physical product 258 | (including a physical distribution medium), accompanied by a 259 | written offer, valid for at least three years and valid for as 260 | long as you offer spare parts or customer support for that product 261 | model, to give anyone who possesses the object code either (1) a 262 | copy of the Corresponding Source for all the software in the 263 | product that is covered by this License, on a durable physical 264 | medium customarily used for software interchange, for a price no 265 | more than your reasonable cost of physically performing this 266 | conveying of source, or (2) access to copy the 267 | Corresponding Source from a network server at no charge. 268 | 269 | c) Convey individual copies of the object code with a copy of the 270 | written offer to provide the Corresponding Source. This 271 | alternative is allowed only occasionally and noncommercially, and 272 | only if you received the object code with such an offer, in accord 273 | with subsection 6b. 274 | 275 | d) Convey the object code by offering access from a designated 276 | place (gratis or for a charge), and offer equivalent access to the 277 | Corresponding Source in the same way through the same place at no 278 | further charge. You need not require recipients to copy the 279 | Corresponding Source along with the object code. If the place to 280 | copy the object code is a network server, the Corresponding Source 281 | may be on a different server (operated by you or a third party) 282 | that supports equivalent copying facilities, provided you maintain 283 | clear directions next to the object code saying where to find the 284 | Corresponding Source. Regardless of what server hosts the 285 | Corresponding Source, you remain obligated to ensure that it is 286 | available for as long as needed to satisfy these requirements. 287 | 288 | e) Convey the object code using peer-to-peer transmission, provided 289 | you inform other peers where the object code and Corresponding 290 | Source of the work are being offered to the general public at no 291 | charge under subsection 6d. 292 | 293 | A separable portion of the object code, whose source code is excluded 294 | from the Corresponding Source as a System Library, need not be 295 | included in conveying the object code work. 296 | 297 | A "User Product" is either (1) a "consumer product", which means any 298 | tangible personal property which is normally used for personal, family, 299 | or household purposes, or (2) anything designed or sold for incorporation 300 | into a dwelling. In determining whether a product is a consumer product, 301 | doubtful cases shall be resolved in favor of coverage. For a particular 302 | product received by a particular user, "normally used" refers to a 303 | typical or common use of that class of product, regardless of the status 304 | of the particular user or of the way in which the particular user 305 | actually uses, or expects or is expected to use, the product. A product 306 | is a consumer product regardless of whether the product has substantial 307 | commercial, industrial or non-consumer uses, unless such uses represent 308 | the only significant mode of use of the product. 309 | 310 | "Installation Information" for a User Product means any methods, 311 | procedures, authorization keys, or other information required to install 312 | and execute modified versions of a covered work in that User Product from 313 | a modified version of its Corresponding Source. The information must 314 | suffice to ensure that the continued functioning of the modified object 315 | code is in no case prevented or interfered with solely because 316 | modification has been made. 317 | 318 | If you convey an object code work under this section in, or with, or 319 | specifically for use in, a User Product, and the conveying occurs as 320 | part of a transaction in which the right of possession and use of the 321 | User Product is transferred to the recipient in perpetuity or for a 322 | fixed term (regardless of how the transaction is characterized), the 323 | Corresponding Source conveyed under this section must be accompanied 324 | by the Installation Information. But this requirement does not apply 325 | if neither you nor any third party retains the ability to install 326 | modified object code on the User Product (for example, the work has 327 | been installed in ROM). 328 | 329 | The requirement to provide Installation Information does not include a 330 | requirement to continue to provide support service, warranty, or updates 331 | for a work that has been modified or installed by the recipient, or for 332 | the User Product in which it has been modified or installed. Access to a 333 | network may be denied when the modification itself materially and 334 | adversely affects the operation of the network or violates the rules and 335 | protocols for communication across the network. 336 | 337 | Corresponding Source conveyed, and Installation Information provided, 338 | in accord with this section must be in a format that is publicly 339 | documented (and with an implementation available to the public in 340 | source code form), and must require no special password or key for 341 | unpacking, reading or copying. 342 | 343 | 7. Additional Terms. 344 | 345 | "Additional permissions" are terms that supplement the terms of this 346 | License by making exceptions from one or more of its conditions. 347 | Additional permissions that are applicable to the entire Program shall 348 | be treated as though they were included in this License, to the extent 349 | that they are valid under applicable law. If additional permissions 350 | apply only to part of the Program, that part may be used separately 351 | under those permissions, but the entire Program remains governed by 352 | this License without regard to the additional permissions. 353 | 354 | When you convey a copy of a covered work, you may at your option 355 | remove any additional permissions from that copy, or from any part of 356 | it. (Additional permissions may be written to require their own 357 | removal in certain cases when you modify the work.) You may place 358 | additional permissions on material, added by you to a covered work, 359 | for which you have or can give appropriate copyright permission. 360 | 361 | Notwithstanding any other provision of this License, for material you 362 | add to a covered work, you may (if authorized by the copyright holders of 363 | that material) supplement the terms of this License with terms: 364 | 365 | a) Disclaiming warranty or limiting liability differently from the 366 | terms of sections 15 and 16 of this License; or 367 | 368 | b) Requiring preservation of specified reasonable legal notices or 369 | author attributions in that material or in the Appropriate Legal 370 | Notices displayed by works containing it; or 371 | 372 | c) Prohibiting misrepresentation of the origin of that material, or 373 | requiring that modified versions of such material be marked in 374 | reasonable ways as different from the original version; or 375 | 376 | d) Limiting the use for publicity purposes of names of licensors or 377 | authors of the material; or 378 | 379 | e) Declining to grant rights under trademark law for use of some 380 | trade names, trademarks, or service marks; or 381 | 382 | f) Requiring indemnification of licensors and authors of that 383 | material by anyone who conveys the material (or modified versions of 384 | it) with contractual assumptions of liability to the recipient, for 385 | any liability that these contractual assumptions directly impose on 386 | those licensors and authors. 387 | 388 | All other non-permissive additional terms are considered "further 389 | restrictions" within the meaning of section 10. If the Program as you 390 | received it, or any part of it, contains a notice stating that it is 391 | governed by this License along with a term that is a further 392 | restriction, you may remove that term. If a license document contains 393 | a further restriction but permits relicensing or conveying under this 394 | License, you may add to a covered work material governed by the terms 395 | of that license document, provided that the further restriction does 396 | not survive such relicensing or conveying. 397 | 398 | If you add terms to a covered work in accord with this section, you 399 | must place, in the relevant source files, a statement of the 400 | additional terms that apply to those files, or a notice indicating 401 | where to find the applicable terms. 402 | 403 | Additional terms, permissive or non-permissive, may be stated in the 404 | form of a separately written license, or stated as exceptions; 405 | the above requirements apply either way. 406 | 407 | 8. Termination. 408 | 409 | You may not propagate or modify a covered work except as expressly 410 | provided under this License. Any attempt otherwise to propagate or 411 | modify it is void, and will automatically terminate your rights under 412 | this License (including any patent licenses granted under the third 413 | paragraph of section 11). 414 | 415 | However, if you cease all violation of this License, then your 416 | license from a particular copyright holder is reinstated (a) 417 | provisionally, unless and until the copyright holder explicitly and 418 | finally terminates your license, and (b) permanently, if the copyright 419 | holder fails to notify you of the violation by some reasonable means 420 | prior to 60 days after the cessation. 421 | 422 | Moreover, your license from a particular copyright holder is 423 | reinstated permanently if the copyright holder notifies you of the 424 | violation by some reasonable means, this is the first time you have 425 | received notice of violation of this License (for any work) from that 426 | copyright holder, and you cure the violation prior to 30 days after 427 | your receipt of the notice. 428 | 429 | Termination of your rights under this section does not terminate the 430 | licenses of parties who have received copies or rights from you under 431 | this License. If your rights have been terminated and not permanently 432 | reinstated, you do not qualify to receive new licenses for the same 433 | material under section 10. 434 | 435 | 9. Acceptance Not Required for Having Copies. 436 | 437 | You are not required to accept this License in order to receive or 438 | run a copy of the Program. Ancillary propagation of a covered work 439 | occurring solely as a consequence of using peer-to-peer transmission 440 | to receive a copy likewise does not require acceptance. However, 441 | nothing other than this License grants you permission to propagate or 442 | modify any covered work. These actions infringe copyright if you do 443 | not accept this License. Therefore, by modifying or propagating a 444 | covered work, you indicate your acceptance of this License to do so. 445 | 446 | 10. Automatic Licensing of Downstream Recipients. 447 | 448 | Each time you convey a covered work, the recipient automatically 449 | receives a license from the original licensors, to run, modify and 450 | propagate that work, subject to this License. You are not responsible 451 | for enforcing compliance by third parties with this License. 452 | 453 | An "entity transaction" is a transaction transferring control of an 454 | organization, or substantially all assets of one, or subdividing an 455 | organization, or merging organizations. If propagation of a covered 456 | work results from an entity transaction, each party to that 457 | transaction who receives a copy of the work also receives whatever 458 | licenses to the work the party's predecessor in interest had or could 459 | give under the previous paragraph, plus a right to possession of the 460 | Corresponding Source of the work from the predecessor in interest, if 461 | the predecessor has it or can get it with reasonable efforts. 462 | 463 | You may not impose any further restrictions on the exercise of the 464 | rights granted or affirmed under this License. For example, you may 465 | not impose a license fee, royalty, or other charge for exercise of 466 | rights granted under this License, and you may not initiate litigation 467 | (including a cross-claim or counterclaim in a lawsuit) alleging that 468 | any patent claim is infringed by making, using, selling, offering for 469 | sale, or importing the Program or any portion of it. 470 | 471 | 11. Patents. 472 | 473 | A "contributor" is a copyright holder who authorizes use under this 474 | License of the Program or a work on which the Program is based. The 475 | work thus licensed is called the contributor's "contributor version". 476 | 477 | A contributor's "essential patent claims" are all patent claims 478 | owned or controlled by the contributor, whether already acquired or 479 | hereafter acquired, that would be infringed by some manner, permitted 480 | by this License, of making, using, or selling its contributor version, 481 | but do not include claims that would be infringed only as a 482 | consequence of further modification of the contributor version. For 483 | purposes of this definition, "control" includes the right to grant 484 | patent sublicenses in a manner consistent with the requirements of 485 | this License. 486 | 487 | Each contributor grants you a non-exclusive, worldwide, royalty-free 488 | patent license under the contributor's essential patent claims, to 489 | make, use, sell, offer for sale, import and otherwise run, modify and 490 | propagate the contents of its contributor version. 491 | 492 | In the following three paragraphs, a "patent license" is any express 493 | agreement or commitment, however denominated, not to enforce a patent 494 | (such as an express permission to practice a patent or covenant not to 495 | sue for patent infringement). To "grant" such a patent license to a 496 | party means to make such an agreement or commitment not to enforce a 497 | patent against the party. 498 | 499 | If you convey a covered work, knowingly relying on a patent license, 500 | and the Corresponding Source of the work is not available for anyone 501 | to copy, free of charge and under the terms of this License, through a 502 | publicly available network server or other readily accessible means, 503 | then you must either (1) cause the Corresponding Source to be so 504 | available, or (2) arrange to deprive yourself of the benefit of the 505 | patent license for this particular work, or (3) arrange, in a manner 506 | consistent with the requirements of this License, to extend the patent 507 | license to downstream recipients. "Knowingly relying" means you have 508 | actual knowledge that, but for the patent license, your conveying the 509 | covered work in a country, or your recipient's use of the covered work 510 | in a country, would infringe one or more identifiable patents in that 511 | country that you have reason to believe are valid. 512 | 513 | If, pursuant to or in connection with a single transaction or 514 | arrangement, you convey, or propagate by procuring conveyance of, a 515 | covered work, and grant a patent license to some of the parties 516 | receiving the covered work authorizing them to use, propagate, modify 517 | or convey a specific copy of the covered work, then the patent license 518 | you grant is automatically extended to all recipients of the covered 519 | work and works based on it. 520 | 521 | A patent license is "discriminatory" if it does not include within 522 | the scope of its coverage, prohibits the exercise of, or is 523 | conditioned on the non-exercise of one or more of the rights that are 524 | specifically granted under this License. You may not convey a covered 525 | work if you are a party to an arrangement with a third party that is 526 | in the business of distributing software, under which you make payment 527 | to the third party based on the extent of your activity of conveying 528 | the work, and under which the third party grants, to any of the 529 | parties who would receive the covered work from you, a discriminatory 530 | patent license (a) in connection with copies of the covered work 531 | conveyed by you (or copies made from those copies), or (b) primarily 532 | for and in connection with specific products or compilations that 533 | contain the covered work, unless you entered into that arrangement, 534 | or that patent license was granted, prior to 28 March 2007. 535 | 536 | Nothing in this License shall be construed as excluding or limiting 537 | any implied license or other defenses to infringement that may 538 | otherwise be available to you under applicable patent law. 539 | 540 | 12. No Surrender of Others' Freedom. 541 | 542 | If conditions are imposed on you (whether by court order, agreement or 543 | otherwise) that contradict the conditions of this License, they do not 544 | excuse you from the conditions of this License. If you cannot convey a 545 | covered work so as to satisfy simultaneously your obligations under this 546 | License and any other pertinent obligations, then as a consequence you may 547 | not convey it at all. For example, if you agree to terms that obligate you 548 | to collect a royalty for further conveying from those to whom you convey 549 | the Program, the only way you could satisfy both those terms and this 550 | License would be to refrain entirely from conveying the Program. 551 | 552 | 13. Use with the GNU Affero General Public License. 553 | 554 | Notwithstanding any other provision of this License, you have 555 | permission to link or combine any covered work with a work licensed 556 | under version 3 of the GNU Affero General Public License into a single 557 | combined work, and to convey the resulting work. The terms of this 558 | License will continue to apply to the part which is the covered work, 559 | but the special requirements of the GNU Affero General Public License, 560 | section 13, concerning interaction through a network will apply to the 561 | combination as such. 562 | 563 | 14. Revised Versions of this License. 564 | 565 | The Free Software Foundation may publish revised and/or new versions of 566 | the GNU General Public License from time to time. Such new versions will 567 | be similar in spirit to the present version, but may differ in detail to 568 | address new problems or concerns. 569 | 570 | Each version is given a distinguishing version number. If the 571 | Program specifies that a certain numbered version of the GNU General 572 | Public License "or any later version" applies to it, you have the 573 | option of following the terms and conditions either of that numbered 574 | version or of any later version published by the Free Software 575 | Foundation. If the Program does not specify a version number of the 576 | GNU General Public License, you may choose any version ever published 577 | by the Free Software Foundation. 578 | 579 | If the Program specifies that a proxy can decide which future 580 | versions of the GNU General Public License can be used, that proxy's 581 | public statement of acceptance of a version permanently authorizes you 582 | to choose that version for the Program. 583 | 584 | Later license versions may give you additional or different 585 | permissions. However, no additional obligations are imposed on any 586 | author or copyright holder as a result of your choosing to follow a 587 | later version. 588 | 589 | 15. Disclaimer of Warranty. 590 | 591 | THERE IS NO WARRANTY FOR THE PROGRAM, TO THE EXTENT PERMITTED BY 592 | APPLICABLE LAW. EXCEPT WHEN OTHERWISE STATED IN WRITING THE COPYRIGHT 593 | HOLDERS AND/OR OTHER PARTIES PROVIDE THE PROGRAM "AS IS" WITHOUT WARRANTY 594 | OF ANY KIND, EITHER EXPRESSED OR IMPLIED, INCLUDING, BUT NOT LIMITED TO, 595 | THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR 596 | PURPOSE. THE ENTIRE RISK AS TO THE QUALITY AND PERFORMANCE OF THE PROGRAM 597 | IS WITH YOU. SHOULD THE PROGRAM PROVE DEFECTIVE, YOU ASSUME THE COST OF 598 | ALL NECESSARY SERVICING, REPAIR OR CORRECTION. 599 | 600 | 16. Limitation of Liability. 601 | 602 | IN NO EVENT UNLESS REQUIRED BY APPLICABLE LAW OR AGREED TO IN WRITING 603 | WILL ANY COPYRIGHT HOLDER, OR ANY OTHER PARTY WHO MODIFIES AND/OR CONVEYS 604 | THE PROGRAM AS PERMITTED ABOVE, BE LIABLE TO YOU FOR DAMAGES, INCLUDING ANY 605 | GENERAL, SPECIAL, INCIDENTAL OR CONSEQUENTIAL DAMAGES ARISING OUT OF THE 606 | USE OR INABILITY TO USE THE PROGRAM (INCLUDING BUT NOT LIMITED TO LOSS OF 607 | DATA OR DATA BEING RENDERED INACCURATE OR LOSSES SUSTAINED BY YOU OR THIRD 608 | PARTIES OR A FAILURE OF THE PROGRAM TO OPERATE WITH ANY OTHER PROGRAMS), 609 | EVEN IF SUCH HOLDER OR OTHER PARTY HAS BEEN ADVISED OF THE POSSIBILITY OF 610 | SUCH DAMAGES. 611 | 612 | 17. Interpretation of Sections 15 and 16. 613 | 614 | If the disclaimer of warranty and limitation of liability provided 615 | above cannot be given local legal effect according to their terms, 616 | reviewing courts shall apply local law that most closely approximates 617 | an absolute waiver of all civil liability in connection with the 618 | Program, unless a warranty or assumption of liability accompanies a 619 | copy of the Program in return for a fee. 620 | 621 | END OF TERMS AND CONDITIONS 622 | 623 | How to Apply These Terms to Your New Programs 624 | 625 | If you develop a new program, and you want it to be of the greatest 626 | possible use to the public, the best way to achieve this is to make it 627 | free software which everyone can redistribute and change under these terms. 628 | 629 | To do so, attach the following notices to the program. It is safest 630 | to attach them to the start of each source file to most effectively 631 | state the exclusion of warranty; and each file should have at least 632 | the "copyright" line and a pointer to where the full notice is found. 633 | 634 | 635 | Copyright (C) 636 | 637 | This program is free software: you can redistribute it and/or modify 638 | it under the terms of the GNU General Public License as published by 639 | the Free Software Foundation, either version 3 of the License, or 640 | (at your option) any later version. 641 | 642 | This program is distributed in the hope that it will be useful, 643 | but WITHOUT ANY WARRANTY; without even the implied warranty of 644 | MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the 645 | GNU General Public License for more details. 646 | 647 | You should have received a copy of the GNU General Public License 648 | along with this program. If not, see . 649 | 650 | Also add information on how to contact you by electronic and paper mail. 651 | 652 | If the program does terminal interaction, make it output a short 653 | notice like this when it starts in an interactive mode: 654 | 655 | Copyright (C) 656 | This program comes with ABSOLUTELY NO WARRANTY; for details type `show w'. 657 | This is free software, and you are welcome to redistribute it 658 | under certain conditions; type `show c' for details. 659 | 660 | The hypothetical commands `show w' and `show c' should show the appropriate 661 | parts of the General Public License. Of course, your program's commands 662 | might be different; for a GUI interface, you would use an "about box". 663 | 664 | You should also get your employer (if you work as a programmer) or school, 665 | if any, to sign a "copyright disclaimer" for the program, if necessary. 666 | For more information on this, and how to apply and follow the GNU GPL, see 667 | . 668 | 669 | The GNU General Public License does not permit incorporating your program 670 | into proprietary programs. If your program is a subroutine library, you 671 | may consider it more useful to permit linking proprietary applications with 672 | the library. If this is what you want to do, use the GNU Lesser General 673 | Public License instead of this License. But first, please read 674 | . 675 | -------------------------------------------------------------------------------- /retrain.py: -------------------------------------------------------------------------------- 1 | # Copyright 2015 The TensorFlow Authors. All Rights Reserved. 2 | # 3 | # Licensed under the Apache License, Version 2.0 (the "License"); 4 | # you may not use this file except in compliance with the License. 5 | # You may obtain a copy of the License at 6 | # 7 | # http://www.apache.org/licenses/LICENSE-2.0 8 | # 9 | # Unless required by applicable law or agreed to in writing, software 10 | # distributed under the License is distributed on an "AS IS" BASIS, 11 | # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. 12 | # See the License for the specific language governing permissions and 13 | # limitations under the License. 14 | # ============================================================================== 15 | """Simple transfer learning with an Inception v3 architecture model which 16 | displays summaries in TensorBoard. 17 | 18 | This example shows how to take a Inception v3 architecture model trained on 19 | ImageNet images, and train a new top layer that can recognize other classes of 20 | images. 21 | 22 | The top layer receives as input a 2048-dimensional vector for each image. We 23 | train a softmax layer on top of this representation. Assuming the softmax layer 24 | contains N labels, this corresponds to learning N + 2048*N model parameters 25 | corresponding to the learned biases and weights. 26 | 27 | Here's an example, which assumes you have a folder containing class-named 28 | subfolders, each full of images for each label. The example folder flower_photos 29 | should have a structure like this: 30 | 31 | ~/flower_photos/daisy/photo1.jpg 32 | ~/flower_photos/daisy/photo2.jpg 33 | ... 34 | ~/flower_photos/rose/anotherphoto77.jpg 35 | ... 36 | ~/flower_photos/sunflower/somepicture.jpg 37 | 38 | The subfolder names are important, since they define what label is applied to 39 | each image, but the filenames themselves don't matter. Once your images are 40 | prepared, you can run the training with a command like this: 41 | 42 | bazel build tensorflow/examples/image_retraining:retrain && \ 43 | bazel-bin/tensorflow/examples/image_retraining/retrain \ 44 | --image_dir ~/flower_photos 45 | 46 | You can replace the image_dir argument with any folder containing subfolders of 47 | images. The label for each image is taken from the name of the subfolder it's 48 | in. 49 | 50 | This produces a new model file that can be loaded and run by any TensorFlow 51 | program, for example the label_image sample code. 52 | 53 | 54 | To use with TensorBoard: 55 | 56 | By default, this script will log summaries to /tmp/retrain_logs directory 57 | 58 | Visualize the summaries with this command: 59 | 60 | tensorboard --logdir /tmp/retrain_logs 61 | 62 | """ 63 | from __future__ import absolute_import 64 | from __future__ import division 65 | from __future__ import print_function 66 | 67 | import argparse 68 | from datetime import datetime 69 | import hashlib 70 | import os.path 71 | import random 72 | import re 73 | import struct 74 | import sys 75 | import tarfile 76 | 77 | import numpy as np 78 | from six.moves import urllib 79 | import tensorflow as tf 80 | 81 | from tensorflow.python.framework import graph_util 82 | from tensorflow.python.framework import tensor_shape 83 | from tensorflow.python.platform import gfile 84 | from tensorflow.python.util import compat 85 | 86 | FLAGS = None 87 | 88 | # These are all parameters that are tied to the particular model architecture 89 | # we're using for Inception v3. These include things like tensor names and their 90 | # sizes. If you want to adapt this script to work with another model, you will 91 | # need to update these to reflect the values in the network you're using. 92 | # pylint: disable=line-too-long 93 | DATA_URL = 'http://download.tensorflow.org/models/image/imagenet/inception-2015-12-05.tgz' 94 | # pylint: enable=line-too-long 95 | BOTTLENECK_TENSOR_NAME = 'pool_3/_reshape:0' 96 | BOTTLENECK_TENSOR_SIZE = 2048 97 | MODEL_INPUT_WIDTH = 299 98 | MODEL_INPUT_HEIGHT = 299 99 | MODEL_INPUT_DEPTH = 3 100 | JPEG_DATA_TENSOR_NAME = 'DecodeJpeg/contents:0' 101 | RESIZED_INPUT_TENSOR_NAME = 'ResizeBilinear:0' 102 | MAX_NUM_IMAGES_PER_CLASS = 2 ** 27 - 1 # ~134M 103 | 104 | 105 | def create_image_lists(image_dir, testing_percentage, validation_percentage): 106 | """Builds a list of training images from the file system. 107 | 108 | Analyzes the sub folders in the image directory, splits them into stable 109 | training, testing, and validation sets, and returns a data structure 110 | describing the lists of images for each label and their paths. 111 | 112 | Args: 113 | image_dir: String path to a folder containing subfolders of images. 114 | testing_percentage: Integer percentage of the images to reserve for tests. 115 | validation_percentage: Integer percentage of images reserved for validation. 116 | 117 | Returns: 118 | A dictionary containing an entry for each label subfolder, with images split 119 | into training, testing, and validation sets within each label. 120 | """ 121 | if not gfile.Exists(image_dir): 122 | print("Image directory '" + image_dir + "' not found.") 123 | return None 124 | result = {} 125 | sub_dirs = [x[0] for x in gfile.Walk(image_dir)] 126 | # The root directory comes first, so skip it. 127 | is_root_dir = True 128 | for sub_dir in sub_dirs: 129 | if is_root_dir: 130 | is_root_dir = False 131 | continue 132 | extensions = ['jpg', 'jpeg', 'JPG', 'JPEG'] 133 | file_list = [] 134 | dir_name = os.path.basename(sub_dir) 135 | if dir_name == image_dir: 136 | continue 137 | print("Looking for images in '" + dir_name + "'") 138 | for extension in extensions: 139 | file_glob = os.path.join(image_dir, dir_name, '*.' + extension) 140 | file_list.extend(gfile.Glob(file_glob)) 141 | if not file_list: 142 | print('No files found') 143 | continue 144 | if len(file_list) < 20: 145 | print('WARNING: Folder has less than 20 images, which may cause issues.') 146 | elif len(file_list) > MAX_NUM_IMAGES_PER_CLASS: 147 | print('WARNING: Folder {} has more than {} images. Some images will ' 148 | 'never be selected.'.format(dir_name, MAX_NUM_IMAGES_PER_CLASS)) 149 | label_name = re.sub(r'[^a-z0-9]+', ' ', dir_name.lower()) 150 | training_images = [] 151 | testing_images = [] 152 | validation_images = [] 153 | for file_name in file_list: 154 | base_name = os.path.basename(file_name) 155 | # We want to ignore anything after '_nohash_' in the file name when 156 | # deciding which set to put an image in, the data set creator has a way of 157 | # grouping photos that are close variations of each other. For example 158 | # this is used in the plant disease data set to group multiple pictures of 159 | # the same leaf. 160 | hash_name = re.sub(r'_nohash_.*$', '', file_name) 161 | # This looks a bit magical, but we need to decide whether this file should 162 | # go into the training, testing, or validation sets, and we want to keep 163 | # existing files in the same set even if more files are subsequently 164 | # added. 165 | # To do that, we need a stable way of deciding based on just the file name 166 | # itself, so we do a hash of that and then use that to generate a 167 | # probability value that we use to assign it. 168 | hash_name_hashed = hashlib.sha1(compat.as_bytes(hash_name)).hexdigest() 169 | percentage_hash = ((int(hash_name_hashed, 16) % 170 | (MAX_NUM_IMAGES_PER_CLASS + 1)) * 171 | (100.0 / MAX_NUM_IMAGES_PER_CLASS)) 172 | if percentage_hash < validation_percentage: 173 | validation_images.append(base_name) 174 | elif percentage_hash < (testing_percentage + validation_percentage): 175 | testing_images.append(base_name) 176 | else: 177 | training_images.append(base_name) 178 | result[label_name] = { 179 | 'dir': dir_name, 180 | 'training': training_images, 181 | 'testing': testing_images, 182 | 'validation': validation_images, 183 | } 184 | return result 185 | 186 | 187 | def get_image_path(image_lists, label_name, index, image_dir, category): 188 | """"Returns a path to an image for a label at the given index. 189 | 190 | Args: 191 | image_lists: Dictionary of training images for each label. 192 | label_name: Label string we want to get an image for. 193 | index: Int offset of the image we want. This will be moduloed by the 194 | available number of images for the label, so it can be arbitrarily large. 195 | image_dir: Root folder string of the subfolders containing the training 196 | images. 197 | category: Name string of set to pull images from - training, testing, or 198 | validation. 199 | 200 | Returns: 201 | File system path string to an image that meets the requested parameters. 202 | 203 | """ 204 | if label_name not in image_lists: 205 | tf.logging.fatal('Label does not exist %s.', label_name) 206 | label_lists = image_lists[label_name] 207 | if category not in label_lists: 208 | tf.logging.fatal('Category does not exist %s.', category) 209 | category_list = label_lists[category] 210 | if not category_list: 211 | tf.logging.fatal('Label %s has no images in the category %s.', 212 | label_name, category) 213 | mod_index = index % len(category_list) 214 | base_name = category_list[mod_index] 215 | sub_dir = label_lists['dir'] 216 | full_path = os.path.join(image_dir, sub_dir, base_name) 217 | return full_path 218 | 219 | 220 | def get_bottleneck_path(image_lists, label_name, index, bottleneck_dir, 221 | category): 222 | """"Returns a path to a bottleneck file for a label at the given index. 223 | 224 | Args: 225 | image_lists: Dictionary of training images for each label. 226 | label_name: Label string we want to get an image for. 227 | index: Integer offset of the image we want. This will be moduloed by the 228 | available number of images for the label, so it can be arbitrarily large. 229 | bottleneck_dir: Folder string holding cached files of bottleneck values. 230 | category: Name string of set to pull images from - training, testing, or 231 | validation. 232 | 233 | Returns: 234 | File system path string to an image that meets the requested parameters. 235 | """ 236 | return get_image_path(image_lists, label_name, index, bottleneck_dir, 237 | category) + '.txt' 238 | 239 | 240 | def create_inception_graph(): 241 | """"Creates a graph from saved GraphDef file and returns a Graph object. 242 | 243 | Returns: 244 | Graph holding the trained Inception network, and various tensors we'll be 245 | manipulating. 246 | """ 247 | with tf.Session() as sess: 248 | model_filename = os.path.join( 249 | FLAGS.model_dir, 'classify_image_graph_def.pb') 250 | with gfile.FastGFile(model_filename, 'rb') as f: 251 | graph_def = tf.GraphDef() 252 | graph_def.ParseFromString(f.read()) 253 | bottleneck_tensor, jpeg_data_tensor, resized_input_tensor = ( 254 | tf.import_graph_def(graph_def, name='', return_elements=[ 255 | BOTTLENECK_TENSOR_NAME, JPEG_DATA_TENSOR_NAME, 256 | RESIZED_INPUT_TENSOR_NAME])) 257 | return sess.graph, bottleneck_tensor, jpeg_data_tensor, resized_input_tensor 258 | 259 | 260 | def run_bottleneck_on_image(sess, image_data, image_data_tensor, 261 | bottleneck_tensor): 262 | """Runs inference on an image to extract the 'bottleneck' summary layer. 263 | 264 | Args: 265 | sess: Current active TensorFlow Session. 266 | image_data: String of raw JPEG data. 267 | image_data_tensor: Input data layer in the graph. 268 | bottleneck_tensor: Layer before the final softmax. 269 | 270 | Returns: 271 | Numpy array of bottleneck values. 272 | """ 273 | bottleneck_values = sess.run( 274 | bottleneck_tensor, 275 | {image_data_tensor: image_data}) 276 | bottleneck_values = np.squeeze(bottleneck_values) 277 | return bottleneck_values 278 | 279 | 280 | def maybe_download_and_extract(): 281 | """Download and extract model tar file. 282 | 283 | If the pretrained model we're using doesn't already exist, this function 284 | downloads it from the TensorFlow.org website and unpacks it into a directory. 285 | """ 286 | dest_directory = FLAGS.model_dir 287 | if not os.path.exists(dest_directory): 288 | os.makedirs(dest_directory) 289 | filename = DATA_URL.split('/')[-1] 290 | filepath = os.path.join(dest_directory, filename) 291 | if not os.path.exists(filepath): 292 | 293 | def _progress(count, block_size, total_size): 294 | sys.stdout.write('\r>> Downloading %s %.1f%%' % 295 | (filename, 296 | float(count * block_size) / float(total_size) * 100.0)) 297 | sys.stdout.flush() 298 | 299 | filepath, _ = urllib.request.urlretrieve(DATA_URL, 300 | filepath, 301 | _progress) 302 | print() 303 | statinfo = os.stat(filepath) 304 | print('Successfully downloaded', filename, statinfo.st_size, 'bytes.') 305 | tarfile.open(filepath, 'r:gz').extractall(dest_directory) 306 | 307 | 308 | def ensure_dir_exists(dir_name): 309 | """Makes sure the folder exists on disk. 310 | 311 | Args: 312 | dir_name: Path string to the folder we want to create. 313 | """ 314 | if not os.path.exists(dir_name): 315 | os.makedirs(dir_name) 316 | 317 | 318 | def write_list_of_floats_to_file(list_of_floats , file_path): 319 | """Writes a given list of floats to a binary file. 320 | 321 | Args: 322 | list_of_floats: List of floats we want to write to a file. 323 | file_path: Path to a file where list of floats will be stored. 324 | 325 | """ 326 | 327 | s = struct.pack('d' * BOTTLENECK_TENSOR_SIZE, *list_of_floats) 328 | with open(file_path, 'wb') as f: 329 | f.write(s) 330 | 331 | 332 | def read_list_of_floats_from_file(file_path): 333 | """Reads list of floats from a given file. 334 | 335 | Args: 336 | file_path: Path to a file where list of floats was stored. 337 | Returns: 338 | Array of bottleneck values (list of floats). 339 | 340 | """ 341 | 342 | with open(file_path, 'rb') as f: 343 | s = struct.unpack('d' * BOTTLENECK_TENSOR_SIZE, f.read()) 344 | return list(s) 345 | 346 | 347 | bottleneck_path_2_bottleneck_values = {} 348 | 349 | def create_bottleneck_file(bottleneck_path, image_lists, label_name, index, 350 | image_dir, category, sess, jpeg_data_tensor, bottleneck_tensor): 351 | print('Creating bottleneck at ' + bottleneck_path) 352 | image_path = get_image_path(image_lists, label_name, index, image_dir, category) 353 | if not gfile.Exists(image_path): 354 | tf.logging.fatal('File does not exist %s', image_path) 355 | image_data = gfile.FastGFile(image_path, 'rb').read() 356 | bottleneck_values = run_bottleneck_on_image(sess, image_data, jpeg_data_tensor, bottleneck_tensor) 357 | bottleneck_string = ','.join(str(x) for x in bottleneck_values) 358 | with open(bottleneck_path, 'w') as bottleneck_file: 359 | bottleneck_file.write(bottleneck_string) 360 | 361 | def get_or_create_bottleneck(sess, image_lists, label_name, index, image_dir, 362 | category, bottleneck_dir, jpeg_data_tensor, 363 | bottleneck_tensor): 364 | """Retrieves or calculates bottleneck values for an image. 365 | 366 | If a cached version of the bottleneck data exists on-disk, return that, 367 | otherwise calculate the data and save it to disk for future use. 368 | 369 | Args: 370 | sess: The current active TensorFlow Session. 371 | image_lists: Dictionary of training images for each label. 372 | label_name: Label string we want to get an image for. 373 | index: Integer offset of the image we want. This will be modulo-ed by the 374 | available number of images for the label, so it can be arbitrarily large. 375 | image_dir: Root folder string of the subfolders containing the training 376 | images. 377 | category: Name string of which set to pull images from - training, testing, 378 | or validation. 379 | bottleneck_dir: Folder string holding cached files of bottleneck values. 380 | jpeg_data_tensor: The tensor to feed loaded jpeg data into. 381 | bottleneck_tensor: The output tensor for the bottleneck values. 382 | 383 | Returns: 384 | Numpy array of values produced by the bottleneck layer for the image. 385 | """ 386 | label_lists = image_lists[label_name] 387 | sub_dir = label_lists['dir'] 388 | sub_dir_path = os.path.join(bottleneck_dir, sub_dir) 389 | ensure_dir_exists(sub_dir_path) 390 | bottleneck_path = get_bottleneck_path(image_lists, label_name, index, bottleneck_dir, category) 391 | if not os.path.exists(bottleneck_path): 392 | create_bottleneck_file(bottleneck_path, image_lists, label_name, index, image_dir, category, sess, jpeg_data_tensor, bottleneck_tensor) 393 | with open(bottleneck_path, 'r') as bottleneck_file: 394 | bottleneck_string = bottleneck_file.read() 395 | did_hit_error = False 396 | try: 397 | bottleneck_values = [float(x) for x in bottleneck_string.split(',')] 398 | except: 399 | print("Invalid float found, recreating bottleneck") 400 | did_hit_error = True 401 | if did_hit_error: 402 | create_bottleneck_file(bottleneck_path, image_lists, label_name, index, image_dir, category, sess, jpeg_data_tensor, bottleneck_tensor) 403 | with open(bottleneck_path, 'r') as bottleneck_file: 404 | bottleneck_string = bottleneck_file.read() 405 | # Allow exceptions to propagate here, since they shouldn't happen after a fresh creation 406 | bottleneck_values = [float(x) for x in bottleneck_string.split(',')] 407 | return bottleneck_values 408 | 409 | def cache_bottlenecks(sess, image_lists, image_dir, bottleneck_dir, 410 | jpeg_data_tensor, bottleneck_tensor): 411 | """Ensures all the training, testing, and validation bottlenecks are cached. 412 | 413 | Because we're likely to read the same image multiple times (if there are no 414 | distortions applied during training) it can speed things up a lot if we 415 | calculate the bottleneck layer values once for each image during 416 | preprocessing, and then just read those cached values repeatedly during 417 | training. Here we go through all the images we've found, calculate those 418 | values, and save them off. 419 | 420 | Args: 421 | sess: The current active TensorFlow Session. 422 | image_lists: Dictionary of training images for each label. 423 | image_dir: Root folder string of the subfolders containing the training 424 | images. 425 | bottleneck_dir: Folder string holding cached files of bottleneck values. 426 | jpeg_data_tensor: Input tensor for jpeg data from file. 427 | bottleneck_tensor: The penultimate output layer of the graph. 428 | 429 | Returns: 430 | Nothing. 431 | """ 432 | how_many_bottlenecks = 0 433 | ensure_dir_exists(bottleneck_dir) 434 | for label_name, label_lists in image_lists.items(): 435 | for category in ['training', 'testing', 'validation']: 436 | category_list = label_lists[category] 437 | for index, unused_base_name in enumerate(category_list): 438 | get_or_create_bottleneck(sess, image_lists, label_name, index, 439 | image_dir, category, bottleneck_dir, 440 | jpeg_data_tensor, bottleneck_tensor) 441 | 442 | how_many_bottlenecks += 1 443 | if how_many_bottlenecks % 100 == 0: 444 | print(str(how_many_bottlenecks) + ' bottleneck files created.') 445 | 446 | 447 | def get_random_cached_bottlenecks(sess, image_lists, how_many, category, 448 | bottleneck_dir, image_dir, jpeg_data_tensor, 449 | bottleneck_tensor): 450 | """Retrieves bottleneck values for cached images. 451 | 452 | If no distortions are being applied, this function can retrieve the cached 453 | bottleneck values directly from disk for images. It picks a random set of 454 | images from the specified category. 455 | 456 | Args: 457 | sess: Current TensorFlow Session. 458 | image_lists: Dictionary of training images for each label. 459 | how_many: If positive, a random sample of this size will be chosen. 460 | If negative, all bottlenecks will be retrieved. 461 | category: Name string of which set to pull from - training, testing, or 462 | validation. 463 | bottleneck_dir: Folder string holding cached files of bottleneck values. 464 | image_dir: Root folder string of the subfolders containing the training 465 | images. 466 | jpeg_data_tensor: The layer to feed jpeg image data into. 467 | bottleneck_tensor: The bottleneck output layer of the CNN graph. 468 | 469 | Returns: 470 | List of bottleneck arrays, their corresponding ground truths, and the 471 | relevant filenames. 472 | """ 473 | class_count = len(image_lists.keys()) 474 | bottlenecks = [] 475 | ground_truths = [] 476 | filenames = [] 477 | if how_many >= 0: 478 | # Retrieve a random sample of bottlenecks. 479 | for unused_i in range(how_many): 480 | label_index = random.randrange(class_count) 481 | label_name = list(image_lists.keys())[label_index] 482 | image_index = random.randrange(MAX_NUM_IMAGES_PER_CLASS + 1) 483 | image_name = get_image_path(image_lists, label_name, image_index, 484 | image_dir, category) 485 | bottleneck = get_or_create_bottleneck(sess, image_lists, label_name, 486 | image_index, image_dir, category, 487 | bottleneck_dir, jpeg_data_tensor, 488 | bottleneck_tensor) 489 | ground_truth = np.zeros(class_count, dtype=np.float32) 490 | ground_truth[label_index] = 1.0 491 | bottlenecks.append(bottleneck) 492 | ground_truths.append(ground_truth) 493 | filenames.append(image_name) 494 | else: 495 | # Retrieve all bottlenecks. 496 | for label_index, label_name in enumerate(image_lists.keys()): 497 | for image_index, image_name in enumerate( 498 | image_lists[label_name][category]): 499 | image_name = get_image_path(image_lists, label_name, image_index, 500 | image_dir, category) 501 | bottleneck = get_or_create_bottleneck(sess, image_lists, label_name, 502 | image_index, image_dir, category, 503 | bottleneck_dir, jpeg_data_tensor, 504 | bottleneck_tensor) 505 | ground_truth = np.zeros(class_count, dtype=np.float32) 506 | ground_truth[label_index] = 1.0 507 | bottlenecks.append(bottleneck) 508 | ground_truths.append(ground_truth) 509 | filenames.append(image_name) 510 | return bottlenecks, ground_truths, filenames 511 | 512 | 513 | def get_random_distorted_bottlenecks( 514 | sess, image_lists, how_many, category, image_dir, input_jpeg_tensor, 515 | distorted_image, resized_input_tensor, bottleneck_tensor): 516 | """Retrieves bottleneck values for training images, after distortions. 517 | 518 | If we're training with distortions like crops, scales, or flips, we have to 519 | recalculate the full model for every image, and so we can't use cached 520 | bottleneck values. Instead we find random images for the requested category, 521 | run them through the distortion graph, and then the full graph to get the 522 | bottleneck results for each. 523 | 524 | Args: 525 | sess: Current TensorFlow Session. 526 | image_lists: Dictionary of training images for each label. 527 | how_many: The integer number of bottleneck values to return. 528 | category: Name string of which set of images to fetch - training, testing, 529 | or validation. 530 | image_dir: Root folder string of the subfolders containing the training 531 | images. 532 | input_jpeg_tensor: The input layer we feed the image data to. 533 | distorted_image: The output node of the distortion graph. 534 | resized_input_tensor: The input node of the recognition graph. 535 | bottleneck_tensor: The bottleneck output layer of the CNN graph. 536 | 537 | Returns: 538 | List of bottleneck arrays and their corresponding ground truths. 539 | """ 540 | class_count = len(image_lists.keys()) 541 | bottlenecks = [] 542 | ground_truths = [] 543 | for unused_i in range(how_many): 544 | label_index = random.randrange(class_count) 545 | label_name = list(image_lists.keys())[label_index] 546 | image_index = random.randrange(MAX_NUM_IMAGES_PER_CLASS + 1) 547 | image_path = get_image_path(image_lists, label_name, image_index, image_dir, 548 | category) 549 | if not gfile.Exists(image_path): 550 | tf.logging.fatal('File does not exist %s', image_path) 551 | jpeg_data = gfile.FastGFile(image_path, 'rb').read() 552 | # Note that we materialize the distorted_image_data as a numpy array before 553 | # sending running inference on the image. This involves 2 memory copies and 554 | # might be optimized in other implementations. 555 | distorted_image_data = sess.run(distorted_image, 556 | {input_jpeg_tensor: jpeg_data}) 557 | bottleneck = run_bottleneck_on_image(sess, distorted_image_data, 558 | resized_input_tensor, 559 | bottleneck_tensor) 560 | ground_truth = np.zeros(class_count, dtype=np.float32) 561 | ground_truth[label_index] = 1.0 562 | bottlenecks.append(bottleneck) 563 | ground_truths.append(ground_truth) 564 | return bottlenecks, ground_truths 565 | 566 | 567 | def should_distort_images(flip_left_right, random_crop, random_scale, 568 | random_brightness): 569 | """Whether any distortions are enabled, from the input flags. 570 | 571 | Args: 572 | flip_left_right: Boolean whether to randomly mirror images horizontally. 573 | random_crop: Integer percentage setting the total margin used around the 574 | crop box. 575 | random_scale: Integer percentage of how much to vary the scale by. 576 | random_brightness: Integer range to randomly multiply the pixel values by. 577 | 578 | Returns: 579 | Boolean value indicating whether any distortions should be applied. 580 | """ 581 | return (flip_left_right or (random_crop != 0) or (random_scale != 0) or 582 | (random_brightness != 0)) 583 | 584 | 585 | def add_input_distortions(flip_left_right, random_crop, random_scale, 586 | random_brightness): 587 | """Creates the operations to apply the specified distortions. 588 | 589 | During training it can help to improve the results if we run the images 590 | through simple distortions like crops, scales, and flips. These reflect the 591 | kind of variations we expect in the real world, and so can help train the 592 | model to cope with natural data more effectively. Here we take the supplied 593 | parameters and construct a network of operations to apply them to an image. 594 | 595 | Cropping 596 | ~~~~~~~~ 597 | 598 | Cropping is done by placing a bounding box at a random position in the full 599 | image. The cropping parameter controls the size of that box relative to the 600 | input image. If it's zero, then the box is the same size as the input and no 601 | cropping is performed. If the value is 50%, then the crop box will be half the 602 | width and height of the input. In a diagram it looks like this: 603 | 604 | < width > 605 | +---------------------+ 606 | | | 607 | | width - crop% | 608 | | < > | 609 | | +------+ | 610 | | | | | 611 | | | | | 612 | | | | | 613 | | +------+ | 614 | | | 615 | | | 616 | +---------------------+ 617 | 618 | Scaling 619 | ~~~~~~~ 620 | 621 | Scaling is a lot like cropping, except that the bounding box is always 622 | centered and its size varies randomly within the given range. For example if 623 | the scale percentage is zero, then the bounding box is the same size as the 624 | input and no scaling is applied. If it's 50%, then the bounding box will be in 625 | a random range between half the width and height and full size. 626 | 627 | Args: 628 | flip_left_right: Boolean whether to randomly mirror images horizontally. 629 | random_crop: Integer percentage setting the total margin used around the 630 | crop box. 631 | random_scale: Integer percentage of how much to vary the scale by. 632 | random_brightness: Integer range to randomly multiply the pixel values by. 633 | graph. 634 | 635 | Returns: 636 | The jpeg input layer and the distorted result tensor. 637 | """ 638 | 639 | jpeg_data = tf.placeholder(tf.string, name='DistortJPGInput') 640 | decoded_image = tf.image.decode_jpeg(jpeg_data, channels=MODEL_INPUT_DEPTH) 641 | decoded_image_as_float = tf.cast(decoded_image, dtype=tf.float32) 642 | decoded_image_4d = tf.expand_dims(decoded_image_as_float, 0) 643 | margin_scale = 1.0 + (random_crop / 100.0) 644 | resize_scale = 1.0 + (random_scale / 100.0) 645 | margin_scale_value = tf.constant(margin_scale) 646 | resize_scale_value = tf.random_uniform(tensor_shape.scalar(), 647 | minval=1.0, 648 | maxval=resize_scale) 649 | scale_value = tf.multiply(margin_scale_value, resize_scale_value) 650 | precrop_width = tf.multiply(scale_value, MODEL_INPUT_WIDTH) 651 | precrop_height = tf.multiply(scale_value, MODEL_INPUT_HEIGHT) 652 | precrop_shape = tf.stack([precrop_height, precrop_width]) 653 | precrop_shape_as_int = tf.cast(precrop_shape, dtype=tf.int32) 654 | precropped_image = tf.image.resize_bilinear(decoded_image_4d, 655 | precrop_shape_as_int) 656 | precropped_image_3d = tf.squeeze(precropped_image, squeeze_dims=[0]) 657 | cropped_image = tf.random_crop(precropped_image_3d, 658 | [MODEL_INPUT_HEIGHT, MODEL_INPUT_WIDTH, 659 | MODEL_INPUT_DEPTH]) 660 | if flip_left_right: 661 | flipped_image = tf.image.random_flip_left_right(cropped_image) 662 | else: 663 | flipped_image = cropped_image 664 | brightness_min = 1.0 - (random_brightness / 100.0) 665 | brightness_max = 1.0 + (random_brightness / 100.0) 666 | brightness_value = tf.random_uniform(tensor_shape.scalar(), 667 | minval=brightness_min, 668 | maxval=brightness_max) 669 | brightened_image = tf.multiply(flipped_image, brightness_value) 670 | distort_result = tf.expand_dims(brightened_image, 0, name='DistortResult') 671 | return jpeg_data, distort_result 672 | 673 | 674 | def variable_summaries(var): 675 | """Attach a lot of summaries to a Tensor (for TensorBoard visualization).""" 676 | with tf.name_scope('summaries'): 677 | mean = tf.reduce_mean(var) 678 | tf.summary.scalar('mean', mean) 679 | with tf.name_scope('stddev'): 680 | stddev = tf.sqrt(tf.reduce_mean(tf.square(var - mean))) 681 | tf.summary.scalar('stddev', stddev) 682 | tf.summary.scalar('max', tf.reduce_max(var)) 683 | tf.summary.scalar('min', tf.reduce_min(var)) 684 | tf.summary.histogram('histogram', var) 685 | 686 | 687 | def add_final_training_ops(class_count, final_tensor_name, bottleneck_tensor): 688 | """Adds a new softmax and fully-connected layer for training. 689 | 690 | We need to retrain the top layer to identify our new classes, so this function 691 | adds the right operations to the graph, along with some variables to hold the 692 | weights, and then sets up all the gradients for the backward pass. 693 | 694 | The set up for the softmax and fully-connected layers is based on: 695 | https://tensorflow.org/versions/master/tutorials/mnist/beginners/index.html 696 | 697 | Args: 698 | class_count: Integer of how many categories of things we're trying to 699 | recognize. 700 | final_tensor_name: Name string for the new final node that produces results. 701 | bottleneck_tensor: The output of the main CNN graph. 702 | 703 | Returns: 704 | The tensors for the training and cross entropy results, and tensors for the 705 | bottleneck input and ground truth input. 706 | """ 707 | with tf.name_scope('input'): 708 | bottleneck_input = tf.placeholder_with_default( 709 | bottleneck_tensor, shape=[None, BOTTLENECK_TENSOR_SIZE], 710 | name='BottleneckInputPlaceholder') 711 | 712 | ground_truth_input = tf.placeholder(tf.float32, 713 | [None, class_count], 714 | name='GroundTruthInput') 715 | 716 | # Organizing the following ops as `final_training_ops` so they're easier 717 | # to see in TensorBoard 718 | layer_name = 'final_training_ops' 719 | with tf.name_scope(layer_name): 720 | with tf.name_scope('weights'): 721 | layer_weights = tf.Variable(tf.truncated_normal([BOTTLENECK_TENSOR_SIZE, class_count], stddev=0.001), name='final_weights') 722 | variable_summaries(layer_weights) 723 | with tf.name_scope('biases'): 724 | layer_biases = tf.Variable(tf.zeros([class_count]), name='final_biases') 725 | variable_summaries(layer_biases) 726 | with tf.name_scope('Wx_plus_b'): 727 | logits = tf.matmul(bottleneck_input, layer_weights) + layer_biases 728 | tf.summary.histogram('pre_activations', logits) 729 | 730 | final_tensor = tf.nn.softmax(logits, name=final_tensor_name) 731 | tf.summary.histogram('activations', final_tensor) 732 | 733 | with tf.name_scope('cross_entropy'): 734 | cross_entropy = tf.nn.softmax_cross_entropy_with_logits( 735 | labels=ground_truth_input, logits=logits) 736 | with tf.name_scope('total'): 737 | cross_entropy_mean = tf.reduce_mean(cross_entropy) 738 | tf.summary.scalar('cross_entropy', cross_entropy_mean) 739 | 740 | with tf.name_scope('train'): 741 | train_step = tf.train.GradientDescentOptimizer(FLAGS.learning_rate).minimize( 742 | cross_entropy_mean) 743 | 744 | return (train_step, cross_entropy_mean, bottleneck_input, ground_truth_input, 745 | final_tensor) 746 | 747 | 748 | def add_evaluation_step(result_tensor, ground_truth_tensor): 749 | """Inserts the operations we need to evaluate the accuracy of our results. 750 | 751 | Args: 752 | result_tensor: The new final node that produces results. 753 | ground_truth_tensor: The node we feed ground truth data 754 | into. 755 | 756 | Returns: 757 | Tuple of (evaluation step, prediction). 758 | """ 759 | with tf.name_scope('accuracy'): 760 | with tf.name_scope('correct_prediction'): 761 | prediction = tf.argmax(result_tensor, 1) 762 | correct_prediction = tf.equal( 763 | prediction, tf.argmax(ground_truth_tensor, 1)) 764 | with tf.name_scope('accuracy'): 765 | evaluation_step = tf.reduce_mean(tf.cast(correct_prediction, tf.float32)) 766 | tf.summary.scalar('accuracy', evaluation_step) 767 | return evaluation_step, prediction 768 | 769 | 770 | def main(_): 771 | # Setup the directory we'll write summaries to for TensorBoard 772 | if tf.gfile.Exists(FLAGS.summaries_dir): 773 | tf.gfile.DeleteRecursively(FLAGS.summaries_dir) 774 | tf.gfile.MakeDirs(FLAGS.summaries_dir) 775 | 776 | # Set up the pre-trained graph. 777 | maybe_download_and_extract() 778 | graph, bottleneck_tensor, jpeg_data_tensor, resized_image_tensor = ( 779 | create_inception_graph()) 780 | 781 | # Look at the folder structure, and create lists of all the images. 782 | image_lists = create_image_lists(FLAGS.image_dir, FLAGS.testing_percentage, 783 | FLAGS.validation_percentage) 784 | class_count = len(image_lists.keys()) 785 | if class_count == 0: 786 | print('No valid folders of images found at ' + FLAGS.image_dir) 787 | return -1 788 | if class_count == 1: 789 | print('Only one valid folder of images found at ' + FLAGS.image_dir + 790 | ' - multiple classes are needed for classification.') 791 | return -1 792 | 793 | # See if the command-line flags mean we're applying any distortions. 794 | do_distort_images = should_distort_images( 795 | FLAGS.flip_left_right, FLAGS.random_crop, FLAGS.random_scale, 796 | FLAGS.random_brightness) 797 | sess = tf.Session() 798 | 799 | if do_distort_images: 800 | # We will be applying distortions, so setup the operations we'll need. 801 | distorted_jpeg_data_tensor, distorted_image_tensor = add_input_distortions( 802 | FLAGS.flip_left_right, FLAGS.random_crop, FLAGS.random_scale, 803 | FLAGS.random_brightness) 804 | else: 805 | # We'll make sure we've calculated the 'bottleneck' image summaries and 806 | # cached them on disk. 807 | cache_bottlenecks(sess, image_lists, FLAGS.image_dir, FLAGS.bottleneck_dir, 808 | jpeg_data_tensor, bottleneck_tensor) 809 | 810 | # Add the new layer that we'll be training. 811 | (train_step, cross_entropy, bottleneck_input, ground_truth_input, 812 | final_tensor) = add_final_training_ops(len(image_lists.keys()), 813 | FLAGS.final_tensor_name, 814 | bottleneck_tensor) 815 | 816 | # Create the operations we need to evaluate the accuracy of our new layer. 817 | evaluation_step, prediction = add_evaluation_step( 818 | final_tensor, ground_truth_input) 819 | 820 | # Merge all the summaries and write them out to /tmp/retrain_logs (by default) 821 | merged = tf.summary.merge_all() 822 | train_writer = tf.summary.FileWriter(FLAGS.summaries_dir + '/train', 823 | sess.graph) 824 | validation_writer = tf.summary.FileWriter(FLAGS.summaries_dir + '/validation') 825 | 826 | # Set up all our weights to their initial default values. 827 | init = tf.global_variables_initializer() 828 | sess.run(init) 829 | 830 | # Run the training for as many cycles as requested on the command line. 831 | for i in range(FLAGS.how_many_training_steps): 832 | # Get a batch of input bottleneck values, either calculated fresh every time 833 | # with distortions applied, or from the cache stored on disk. 834 | if do_distort_images: 835 | train_bottlenecks, train_ground_truth = get_random_distorted_bottlenecks( 836 | sess, image_lists, FLAGS.train_batch_size, 'training', 837 | FLAGS.image_dir, distorted_jpeg_data_tensor, 838 | distorted_image_tensor, resized_image_tensor, bottleneck_tensor) 839 | else: 840 | train_bottlenecks, train_ground_truth, _ = get_random_cached_bottlenecks( 841 | sess, image_lists, FLAGS.train_batch_size, 'training', 842 | FLAGS.bottleneck_dir, FLAGS.image_dir, jpeg_data_tensor, 843 | bottleneck_tensor) 844 | # Feed the bottlenecks and ground truth into the graph, and run a training 845 | # step. Capture training summaries for TensorBoard with the `merged` op. 846 | train_summary, _ = sess.run([merged, train_step], 847 | feed_dict={bottleneck_input: train_bottlenecks, 848 | ground_truth_input: train_ground_truth}) 849 | train_writer.add_summary(train_summary, i) 850 | 851 | # Every so often, print out how well the graph is training. 852 | is_last_step = (i + 1 == FLAGS.how_many_training_steps) 853 | if (i % FLAGS.eval_step_interval) == 0 or is_last_step: 854 | train_accuracy, cross_entropy_value = sess.run( 855 | [evaluation_step, cross_entropy], 856 | feed_dict={bottleneck_input: train_bottlenecks, 857 | ground_truth_input: train_ground_truth}) 858 | print('%s: Step %d: Train accuracy = %.1f%%' % (datetime.now(), i, 859 | train_accuracy * 100)) 860 | print('%s: Step %d: Cross entropy = %f' % (datetime.now(), i, 861 | cross_entropy_value)) 862 | validation_bottlenecks, validation_ground_truth, _ = ( 863 | get_random_cached_bottlenecks( 864 | sess, image_lists, FLAGS.validation_batch_size, 'validation', 865 | FLAGS.bottleneck_dir, FLAGS.image_dir, jpeg_data_tensor, 866 | bottleneck_tensor)) 867 | # Run a validation step and capture training summaries for TensorBoard 868 | # with the `merged` op. 869 | validation_summary, validation_accuracy = sess.run( 870 | [merged, evaluation_step], 871 | feed_dict={bottleneck_input: validation_bottlenecks, 872 | ground_truth_input: validation_ground_truth}) 873 | validation_writer.add_summary(validation_summary, i) 874 | print('%s: Step %d: Validation accuracy = %.1f%% (N=%d)' % 875 | (datetime.now(), i, validation_accuracy * 100, 876 | len(validation_bottlenecks))) 877 | 878 | # We've completed all our training, so run a final test evaluation on 879 | # some new images we haven't used before. 880 | test_bottlenecks, test_ground_truth, test_filenames = ( 881 | get_random_cached_bottlenecks(sess, image_lists, FLAGS.test_batch_size, 882 | 'testing', FLAGS.bottleneck_dir, 883 | FLAGS.image_dir, jpeg_data_tensor, 884 | bottleneck_tensor)) 885 | test_accuracy, predictions = sess.run( 886 | [evaluation_step, prediction], 887 | feed_dict={bottleneck_input: test_bottlenecks, 888 | ground_truth_input: test_ground_truth}) 889 | print('Final test accuracy = %.1f%% (N=%d)' % ( 890 | test_accuracy * 100, len(test_bottlenecks))) 891 | 892 | if FLAGS.print_misclassified_test_images: 893 | print('=== MISCLASSIFIED TEST IMAGES ===') 894 | for i, test_filename in enumerate(test_filenames): 895 | if predictions[i] != test_ground_truth[i].argmax(): 896 | print('%70s %s' % (test_filename, 897 | list(image_lists.keys())[predictions[i]])) 898 | 899 | # Write out the trained graph and labels with the weights stored as constants. 900 | output_graph_def = graph_util.convert_variables_to_constants( 901 | sess, graph.as_graph_def(), [FLAGS.final_tensor_name]) 902 | with gfile.FastGFile(FLAGS.output_graph, 'wb') as f: 903 | f.write(output_graph_def.SerializeToString()) 904 | with gfile.FastGFile(FLAGS.output_labels, 'w') as f: 905 | f.write('\n'.join(image_lists.keys()) + '\n') 906 | 907 | 908 | if __name__ == '__main__': 909 | parser = argparse.ArgumentParser() 910 | parser.add_argument( 911 | '--image_dir', 912 | type=str, 913 | default='', 914 | help='Path to folders of labeled images.' 915 | ) 916 | parser.add_argument( 917 | '--output_graph', 918 | type=str, 919 | default='/tmp/output_graph.pb', 920 | help='Where to save the trained graph.' 921 | ) 922 | parser.add_argument( 923 | '--output_labels', 924 | type=str, 925 | default='/tmp/output_labels.txt', 926 | help='Where to save the trained graph\'s labels.' 927 | ) 928 | parser.add_argument( 929 | '--summaries_dir', 930 | type=str, 931 | default='/tmp/retrain_logs', 932 | help='Where to save summary logs for TensorBoard.' 933 | ) 934 | parser.add_argument( 935 | '--how_many_training_steps', 936 | type=int, 937 | default=4000, 938 | help='How many training steps to run before ending.' 939 | ) 940 | parser.add_argument( 941 | '--learning_rate', 942 | type=float, 943 | default=0.01, 944 | help='How large a learning rate to use when training.' 945 | ) 946 | parser.add_argument( 947 | '--testing_percentage', 948 | type=int, 949 | default=10, 950 | help='What percentage of images to use as a test set.' 951 | ) 952 | parser.add_argument( 953 | '--validation_percentage', 954 | type=int, 955 | default=10, 956 | help='What percentage of images to use as a validation set.' 957 | ) 958 | parser.add_argument( 959 | '--eval_step_interval', 960 | type=int, 961 | default=10, 962 | help='How often to evaluate the training results.' 963 | ) 964 | parser.add_argument( 965 | '--train_batch_size', 966 | type=int, 967 | default=100, 968 | help='How many images to train on at a time.' 969 | ) 970 | parser.add_argument( 971 | '--test_batch_size', 972 | type=int, 973 | default=-1, 974 | help="""\ 975 | How many images to test on. This test set is only used once, to evaluate 976 | the final accuracy of the model after training completes. 977 | A value of -1 causes the entire test set to be used, which leads to more 978 | stable results across runs.\ 979 | """ 980 | ) 981 | parser.add_argument( 982 | '--validation_batch_size', 983 | type=int, 984 | default=100, 985 | help="""\ 986 | How many images to use in an evaluation batch. This validation set is 987 | used much more often than the test set, and is an early indicator of how 988 | accurate the model is during training. 989 | A value of -1 causes the entire validation set to be used, which leads to 990 | more stable results across training iterations, but may be slower on large 991 | training sets.\ 992 | """ 993 | ) 994 | parser.add_argument( 995 | '--print_misclassified_test_images', 996 | default=False, 997 | help="""\ 998 | Whether to print out a list of all misclassified test images.\ 999 | """, 1000 | action='store_true' 1001 | ) 1002 | parser.add_argument( 1003 | '--model_dir', 1004 | type=str, 1005 | default='/tmp/imagenet', 1006 | help="""\ 1007 | Path to classify_image_graph_def.pb, 1008 | imagenet_synset_to_human_label_map.txt, and 1009 | imagenet_2012_challenge_label_map_proto.pbtxt.\ 1010 | """ 1011 | ) 1012 | parser.add_argument( 1013 | '--bottleneck_dir', 1014 | type=str, 1015 | default='/tmp/bottleneck', 1016 | help='Path to cache bottleneck layer values as files.' 1017 | ) 1018 | parser.add_argument( 1019 | '--final_tensor_name', 1020 | type=str, 1021 | default='final_result', 1022 | help="""\ 1023 | The name of the output classification layer in the retrained graph.\ 1024 | """ 1025 | ) 1026 | parser.add_argument( 1027 | '--flip_left_right', 1028 | default=False, 1029 | help="""\ 1030 | Whether to randomly flip half of the training images horizontally.\ 1031 | """, 1032 | action='store_true' 1033 | ) 1034 | parser.add_argument( 1035 | '--random_crop', 1036 | type=int, 1037 | default=0, 1038 | help="""\ 1039 | A percentage determining how much of a margin to randomly crop off the 1040 | training images.\ 1041 | """ 1042 | ) 1043 | parser.add_argument( 1044 | '--random_scale', 1045 | type=int, 1046 | default=0, 1047 | help="""\ 1048 | A percentage determining how much to randomly scale up the size of the 1049 | training images by.\ 1050 | """ 1051 | ) 1052 | parser.add_argument( 1053 | '--random_brightness', 1054 | type=int, 1055 | default=0, 1056 | help="""\ 1057 | A percentage determining how much to randomly multiply the training image 1058 | input pixels up or down by.\ 1059 | """ 1060 | ) 1061 | FLAGS, unparsed = parser.parse_known_args() 1062 | tf.app.run(main=main, argv=[sys.argv[0]] + unparsed) 1063 | --------------------------------------------------------------------------------