Awesome

Recurrently Aggregating Deep Features for Salient Object Detection

by Xiaowei Hu, Lei Zhu, Jing Qin, Chi-Wing Fu, and Pheng-Ann Heng

This implementation is written by Xiaowei Hu at the Chinese University of Hong Kong.

Citation

@inproceedings{hu18recurrently,   
  author = {Hu, Xiaowei and Zhu, Lei and Qin, Jing and Fu, Chi-Wing and Heng, Pheng-Ann},    
  title = {Recurrently Aggregating Deep Features for Salient Object Detection},    
  booktitle = {AAAI},    
  pages={6943--6950},         
  year  = {2018}    
}

Saliency Maps

The results of salient object detection on five datasets (ECSSD, HKU-IS, PASCAL-S, SOD, DUT-OMRON) can be found at Google Drive.

Installation

*This model is tested on Ubuntu 16.04, CUDA 8.0, cuDNN 5.0

Please download and compile our CF-Caffe.
Clone the RADF repository, and we'll call the directory that you cloned as RADF-master.
```
git clone https://github.com/xw-hu/RADF.git
```
Replace CF-Caffe/examples/ by RADF-master/examples/. Replace CF-Caffe/data/ by RADF-master/data/.

Test

Put the pretrained model in examples/RADF/snapshot/.
Export PYTHONPATH in the command window such as:
```
export PYTHONPATH='/path/to/CF-Caffe/python'
```
Make the folder examples/RADF/result/ and run the test model in examples/RADF/ (please modify the path of images):
```
ipython notebook RADF_test.ipynb
```
Apply CRF to do the post-processing for each image.
The code for CRF can be found in https://github.com/Andrew-Qibin/dss_crf
*Note that please provide a link to the original code as a footnote or a citation if you plan to use it.

Train

Download the pre-trained VGG16 model at http://www.robots.ox.ac.uk/~vgg/research/very_deep/.
Put this model in CF-Caffe/models/
Enter the examples/RADF/
Modify the image path in train_val.prototxt.
Run
```
sh train.sh
```

Useful Links

*Note that please provide a link to the original code as a footnote or a citation if you plan to use it.