├── IIAU2018.md ├── IIAU_saliency2018 └── README.md /IIAU2018.md: -------------------------------------------------------------------------------- 1 | 2 | IIAU (2018) Saliency Detection Models 3 | 4 | Introduction 5 | ------------------------------------------------------------------------------------------------------------------ 6 | This file contains our saliency detection models appeared in 2018. 7 | 1) LFR (IJCAI2018) 8 | 9 | Title: Salient Object Detection by Lossless Feature Reflection 10 | 11 | Authors: Pingping Zhang, Wei Liu, Huchuan Lu* and Chunhua Shen 12 | 13 | Paper Links: https://arxiv.org/abs/1802.06527 14 | 15 | 2) HyperFusion-Net 16 | 17 | Title: HyperFusion-Net: Densely Reflective Fusion for Salient Object Detection 18 | 19 | Authors: Pingping Zhang, Huchuan Lu* and Chunhua Shen 20 | 21 | Paper Links: https://arxiv.org/abs/1804.05142 22 | 23 | 3) Agile Amulet 24 | 25 | Title: Agile Amulet: Real-Time Salient Object Detection with Contextual Attention 26 | 27 | Authors: Pingping Zhang, Luyao Wang, Dong Wang, Huchuan Lu* and Chunhua Shen 28 | 29 | Paper Links: https://arxiv.org/abs/1804.05142 30 | 31 | How to use 32 | -------------------------------------------------------------------------------------------------------------- 33 | 34 | Prerequisites: 35 | 36 | Download source code from Link: https://1drv.ms/u/s!AnzjJimFPB2IcMN1agL18fD18Us 37 | 38 | Follow the official websites of the Caffe framework and install the whole toolbox (necessary Matlab wrappers) 39 | 40 | 1) Training 41 | 42 | The training code is in ./models/T-Net,./models/HDFN and ./models/AAmulet. For saving memory, we use the BN method implemented by Alex Kendall etal http://mi.eng.cam.ac.uk/projects/segnet/. The training is followed as examples in Caffe. 43 | We use the ImageData layer to load training image pairs. The data list can be generated by Matlab in 44 | 45 | Google Drive Link: https://drive.google.com/open?id=0B5t8yubOMmTKblFZajBuTTYyanc 46 | 47 | Baidu Pan Link: http://pan.baidu.com/s/1pKXnBdD 48 | 49 | The only need is changing the path of data files. 50 | 51 | 2) Testing 52 | 53 | The testing code is in ./matlab/. 54 | 55 | For saliency testing, 56 | 57 | (1) Get prediction : test_saliency_dataset.m 58 | 59 | (2) Get PR value : PR.m 60 | 61 | (3) Plot PR curves: code_pr.m 62 | 63 | (4) Plot Bar figures: code_bar.m 64 | 65 | (5) Get MAE for each method : getmae.m 66 | 67 | Note that we have provided the PR curves in ./PR_curves 68 | 69 | Results 70 | --------------------------------------------------------------------------------------------------------------------- 71 | The saliency maps can be download via baidu yun at http://pan.baidu.com/s/1o7O3lQq. 72 | 73 | Citation 74 | --------------------------------------------------------------------------------------------------------------------- 75 | Please cite the following papers if our models help your research: 76 | 77 | @article{zhang2018salient, 78 | 79 | title={Salient Object Detection by Lossless Feature Reflection}, 80 | 81 | author={Zhang, Pingping and Liu, Wei and Lu, Huchuan and Shen, Chunhua}, 82 | 83 | journal={arXiv preprint arXiv:1802.06527}, 84 | 85 | year={2018} 86 | 87 | } 88 | 89 | @article{zhang2018hyperfusion, 90 | 91 | title={HyperFusion-Net: Densely Reflective Fusion for Salient Object Detection}, 92 | 93 | author={Zhang, Pingping and Lu, Huchuan and Shen, Chunhua}, 94 | 95 | journal={arXiv preprint arXiv:1804.05142}, 96 | 97 | year={2018} 98 | 99 | } 100 | 101 | @article{zhang2018agile, 102 | 103 | title={Agile Amulet: Real-Time Salient Object Detection with Contextual Attention}, 104 | 105 | author={Zhang, Pingping and Wang, Luyao and Wang, Dong and Lu, Huchuan and Shen, Chunhua}, 106 | 107 | journal={arXiv preprint arXiv:1802.06960}, 108 | 109 | year={2018} 110 | 111 | } 112 | 113 | Question and connection 114 | ------------------------------------------------------------------------------------------------------------------ 115 | If any question, please connect 116 | 117 | jssxzhpp@gmail.com; jssxzhpp@mail.dlut.edu.cn 118 | 119 | -------------------------------------------------------------------------------- /IIAU_saliency2018: -------------------------------------------------------------------------------- 1 | 2 | IIAU (2018) Saliency Detection Models 3 | 4 | Introduction 5 | ------------------------------------------------------------------------------------------------------------------ 6 | This file contains our saliency detection models appeared in 2018. 7 | 1) LFR (IJCAI2018) 8 | 9 | Title: Salient Object Detection by Lossless Feature Reflection 10 | 11 | Authors: Pingping Zhang, Wei Liu, Huchuan Lu* and Chunhua Shen 12 | 13 | Paper Links: https://arxiv.org/abs/1802.06527 14 | 15 | 2) HyperFusion-Net 16 | 17 | Title: HyperFusion-Net: Densely Reflective Fusion for Salient Object Detection 18 | 19 | Authors: Pingping Zhang, Huchuan Lu* and Chunhua Shen 20 | 21 | Paper Links: https://arxiv.org/abs/1804.05142 22 | 23 | 3) Agile Amulet 24 | 25 | Title: Agile Amulet: Real-Time Salient Object Detection with Contextual Attention 26 | 27 | Authors: Pingping Zhang, Luyao Wang, Dong Wang, Huchuan Lu* and Chunhua Shen 28 | 29 | Paper Links: https://arxiv.org/abs/1804.05142 30 | 31 | How to use 32 | -------------------------------------------------------------------------------------------------------------- 33 | 34 | Prerequisites: 35 | 36 | Download source code from Link: https://1drv.ms/u/s!AnzjJimFPB2IcMN1agL18fD18Us 37 | 38 | Follow the official websites of the Caffe framework and install the whole toolbox (necessary Matlab wrappers) 39 | 40 | 1) Training 41 | 42 | The training code is in ./models/T-Net,./models/HDFN and ./models/AAmulet. For saving memory, we use the BN method implemented by Alex Kendall etal http://mi.eng.cam.ac.uk/projects/segnet/. The training is followed as examples in Caffe. 43 | We use the ImageData layer to load training image pairs. The data list can be generated by Matlab in 44 | 45 | Google Drive Link: https://drive.google.com/open?id=0B5t8yubOMmTKblFZajBuTTYyanc 46 | 47 | Baidu Pan Link: http://pan.baidu.com/s/1pKXnBdD 48 | 49 | The only need is changing the path of data files. 50 | 51 | 2) Testing 52 | 53 | The testing code is in ./matlab/. 54 | 55 | For saliency testing, 56 | 57 | (1) Get prediction : test_saliency_dataset.m 58 | 59 | (2) Get PR value : PR.m 60 | 61 | (3) Plot PR curves: code_pr.m 62 | 63 | (4) Plot Bar figures: code_bar.m 64 | 65 | (5) Get MAE for each method : getmae.m 66 | 67 | Note that we have provided the PR curves in ./PR_curves 68 | 69 | Citation 70 | --------------------------------------------------------------------------------------------------------------------- 71 | Please cite the following papers if our models help your research: 72 | 73 | @article{zhang2018salient, 74 | 75 | title={Salient Object Detection by Lossless Feature Reflection}, 76 | 77 | author={Zhang, Pingping and Liu, Wei and Lu, Huchuan and Shen, Chunhua}, 78 | 79 | journal={arXiv preprint arXiv:1802.06527}, 80 | 81 | year={2018} 82 | 83 | } 84 | 85 | @article{zhang2018hyperfusion, 86 | 87 | title={HyperFusion-Net: Densely Reflective Fusion for Salient Object Detection}, 88 | 89 | author={Zhang, Pingping and Lu, Huchuan and Shen, Chunhua}, 90 | 91 | journal={arXiv preprint arXiv:1804.05142}, 92 | 93 | year={2018} 94 | 95 | } 96 | 97 | @article{zhang2018agile, 98 | 99 | title={Agile Amulet: Real-Time Salient Object Detection with Contextual Attention}, 100 | 101 | author={Zhang, Pingping and Wang, Luyao and Wang, Dong and Lu, Huchuan and Shen, Chunhua}, 102 | 103 | journal={arXiv preprint arXiv:1802.06960}, 104 | 105 | year={2018} 106 | 107 | } 108 | Question and connection 109 | ------------------------------------------------------------------------------------------------------------------ 110 | If any question, please connect 111 | 112 | jssxzhpp@gmail.com; jssxzhpp@mail.dlut.edu.cn 113 | 114 | -------------------------------------------------------------------------------- /README.md: -------------------------------------------------------------------------------- 1 | IIAU (2017) Saliency Detection Models 2 | 3 | Introduction 4 | ------------------------------------------------------------------------------------------------------------------ 5 | This file contains our saliency detection models appeared in ICCV2017. We delve into the convolutional features in pre-trained FCN models and propose the following methods. 6 | 7 | 1) Amulet 8 | 9 | Title: Amulet: Aggregating Multi-level Convolutional Features for Salient Object Detection 10 | 11 | Authors: Pingping Zhang, Dong Wang, Huchuan Lu*, Hongyu Wang and Xiang Ruan 12 | 13 | Paper Links: 14 | 15 | Google Drive Link: https://drive.google.com/file/d/0B5t8yubOMmTKUnZUYTY5NDdWcTg/view?usp=sharing 16 | 17 | One Drive Link: https://1drv.ms/u/s!AnzjJimFPB2Ibqy-4-sxZOHnb2s 18 | 19 | Baidu Pan Link: https://pan.baidu.com/s/1gfKkIwN 20 | 21 | 2) UCF 22 | 23 | Title: Learning Uncertain Convolutional Features for Accurate Saliency Detection 24 | 25 | Authors: Pingping Zhang, Dong Wang, Huchuan Lu*, Hongyu Wang and Baocai Yin 26 | 27 | Paper Links: 28 | 29 | Google Drive Link: https://drive.google.com/file/d/0B5t8yubOMmTKbUFUQ1gtekVnNEE/view?usp=sharing 30 | 31 | One Drive Link: https://1drv.ms/u/s!AnzjJimFPB2IbY_bYGa-d5HemFw 32 | 33 | Baidu Pan Link: https://pan.baidu.com/s/1skCi4Dv 34 | 35 | 3) SRM 36 | 37 | Title: A Stagewise Refinement Model for Detecting Salient Objects in Images 38 | 39 | Authors: Tiantian Wang, Ali Borji,Lihe Zhang,Pingping Zhang, Huchuan Lu* 40 | 41 | Paper Links: 42 | 43 | Google Drive Link: https://drive.google.com/file/d/0B_MpGgTntG47b18yUlJxOGtuNDA/view 44 | 45 | One Drive Link: https://1drv.ms/u/s!AnzjJimFPB2IbY_bYGa-d5HemFw 46 | 47 | Baidu Pan Link: https://pan.baidu.com/s/1skCi4Dv 48 | 49 | How to use 50 | -------------------------------------------------------------------------------------------------------------- 51 | 52 | Prerequisites: 53 | 54 | Download source code from 55 | 56 | Google Drive Link: https://drive.google.com/open?id=0B5t8yubOMmTKazhHdV8yR19tek0 57 | 58 | One Drive Link: https://1drv.ms/u/s!AnzjJimFPB2Ib_zU3E15C6AxkXQ 59 | 60 | Baidu Pan Link: http://pan.baidu.com/s/1bTAYNc 61 | 62 | 63 | Follow the official websites of the Caffe framework and install the whole toolbox (necessary Matlab wrappers) 64 | 65 | 1) Training 66 | 67 | The training code is in ./models/Amulet and ./models/UCF. For saving memory, we use the BN method implemented by Alex Kendall etal http://mi.eng.cam.ac.uk/projects/segnet/. The training is followed as examples in Caffe. 68 | We use the ImageData layer to load training image pairs. The data list can be generated by Matlab in 69 | 70 | Google Drive Link: https://drive.google.com/open?id=0B5t8yubOMmTKblFZajBuTTYyanc 71 | 72 | Baidu Pan Link: http://pan.baidu.com/s/1pKXnBdD 73 | 74 | The only need is changing the path of data files. 75 | 76 | 2) Testing 77 | 78 | The testing code is in ./matlab/Amulet_test and ./matlab/UCF_test. 79 | 80 | For saliency testing, 81 | 82 | (1) Get prediction : test_saliency_dataset.m 83 | 84 | (2) Get PR value : PR.m 85 | 86 | (3) Plot PR curves: code_pr.m 87 | 88 | (4) Plot Bar figures: code_bar.m 89 | 90 | (5) Get MAE for each method : getmae.m 91 | 92 | Note that we have provided the PR curves in ./PR_curves 93 | 94 | Results 95 | --------------------------------------------------------------------------------------------------------------------- 96 | The saliency maps can be download via baidu yun at http://pan.baidu.com/s/1o7O3lQq. 97 | 98 | Citation 99 | --------------------------------------------------------------------------------------------------------------------- 100 | Please cite the following papers if our models help your research: 101 | 102 | @article{zhang2017amulet, 103 | 104 | title={Amulet: Aggregating Multi-level Convolutional Features for Salient Object Detection}, 105 | 106 | author={Zhang, Pingping and Wang, Dong and Lu, Huchuan and Wang, Hongyu and Ruan, Xiang}, 107 | 108 | journal={arXiv preprint arXiv:1708.02001}, 109 | 110 | year={2017} 111 | 112 | } 113 | 114 | @InProceedings{Zhang_2017_ICCV, 115 | 116 | author = {Zhang, Pingping and Wang, Dong and Lu, Huchuan and Wang, Hongyu and Ruan, Xiang}, 117 | 118 | title = {Amulet: Aggregating Multi-Level Convolutional Features for Salient Object Detection}, 119 | 120 | booktitle = {The IEEE International Conference on Computer Vision (ICCV)}, 121 | 122 | month = {Oct}, 123 | 124 | year = {2017} 125 | 126 | } 127 | 128 | @article{zhang2017learning, 129 | 130 | title={Learning Uncertain Convolutional Features for Accurate Saliency Detection}, 131 | 132 | author={Zhang, Pingping and Wang, Dong and Lu, Huchuan and Wang, Hongyu and Yin, Baocai}, 133 | 134 | journal={arXiv preprint arXiv:1708.02031}, 135 | 136 | year={2017} 137 | 138 | } 139 | 140 | @InProceedings{Zhang_2017_ICCV, 141 | 142 | author = {Zhang, Pingping and Wang, Dong and Lu, Huchuan and Wang, Hongyu and Yin, Baocai}, 143 | 144 | title = {Learning Uncertain Convolutional Features for Accurate Saliency Detection}, 145 | 146 | booktitle = {The IEEE International Conference on Computer Vision (ICCV)}, 147 | 148 | month = {Oct}, 149 | 150 | year = {2017} 151 | 152 | } 153 | 154 | @InProceedings{Wang_2017_ICCV, 155 | 156 | author = {Wang, Tiantian and Borji, Ali and Zhang, Lihe and Zhang, Pingping and Lu, Huchuan}, 157 | 158 | title = {A Stagewise Refinement Model for Detecting Salient Objects in Images}, 159 | 160 | booktitle = {The IEEE International Conference on Computer Vision (ICCV)}, 161 | 162 | month = {Oct}, 163 | 164 | year = {2017} 165 | 166 | } 167 | 168 | Question and connection 169 | ------------------------------------------------------------------------------------------------------------------ 170 | If any question, please connect 171 | 172 | jssxzhpp@gmail.com; jssxzhpp@mail.dlut.edu.cn 173 | --------------------------------------------------------------------------------