/deepsaldet

Source code for our CVPR 2015 work on saliency detection by multi-context deep learning

Primary LanguageC++

  1. build image sudo nvidia-docker build --no-cache -t deepsaldet .
  2. run container sudo nvidia-docker run --name deepsaldet-inst -dit -v /path/to/your/images/dir:/deepsaldet/images deepsaldet
  3. process your images sudo nvidia-docker exec deepsaldet-inst bash /deepsaldet/get_deep_multicontext_saliency.sh

deepsaldet

Source code for our CVPR 2015 work on saliency detection by multi-context deep learning.

Created by Rui Zhao, on May 21, 2015

Summary

This source code is mainly written in Python and bash shell scripts, and it is for the following paper:

  • Rui Zhao, Wanli Ouyang, Hongsheng Li, and Xiaogang Wang. Saliency Detection by Multi-Context Deep Learning. In CVPR 2015.

Usage

  • Supported OS: this source code was tested on 64-bit Arch Linux OS, and it should also be executable in other linux distributions.
  • Pre-installations: refer to caffe for packages required by caffe toolkit. Packages requried by Python scripts include
    • skimage
    • leveldb
    • matplotlib
  • Download caffe models: cd models/ && sh get_models.sh && cd .. (or you can download manually via Baidu Yun: http://pan.baidu.com/s/1sjoP8Ln Password: enn9)
  • Customize test images: put your test images in folder ./images, or revise the test_folder in get_deep_mutlicontext_saliency.sh to your customized image folder.
  • Run demo in bash shell:
        sh get_deep_mutlicontext_saliency.sh

Remark

  • Caffe-sal is a customized version of original caffe toolkit. Comparing the original version, revisions happen in the following files:

    • ./caffe-sal/src/caffe/layers/mcwindowdatalayers.cpp
    • ./caffe-sal/src/caffe/layers/mcwindowdatalayers.cu
    • ./caffe-sal/src/caffe/proto/caffe.proto
    • ./caffe-sal/src/caffe/layer_factory.cpp
    • ./caffe-sal/include/caffe/data_layers.hpp

  • Test folder can be set in ./get_deep_multicontext_saliency.sh

  • This source code requires GPU to accelerate the testing process

  • If everything runs correctly, it will generate resulting saliency maps in test folder (./images), suffix _sc means results produced by single-context deep model, and _mc by multi-context deep model.

##Citing our work Please kindly cite our work in your publications if it helps your research:

@inproceedings{zhao2015saliency,
    title = {Saliency Detection by Multi-Context Deep Learning},
    author={Zhao, Rui and Ouyang, Wanli and Li, Hongsheng and Wang, Xiaogang},
    booktitle = {IEEE Conference on Computer Vision and Pattern
	Recognition (CVPR)},
    year = {2015}
}

##License

Copyright (c) 2015, Rui Zhao
All rights reserved. 

Redistribution and use in source and binary forms, with or without 
modification, are permitted provided that the following conditions are 
met:
		* Redistributions of source code must retain the above copyright 
  		  notice, this list of conditions and the following disclaimer.
		* Redistributions in binary form must reproduce the above copyright 
  		  notice, this list of conditions and the following disclaimer in 
  		  the documentation and/or other materials provided with the distribution

THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" 
AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 
IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 
ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE 	
LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR 
CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF 
SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS 
INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN 
CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) 
ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE 
POSSIBILITY OF SUCH DAMAGE.