sshige/rgbd-saliency

Project page for paper : Learning RGB-D Salient Object Detection using background enclosure, depth contrast, and top-down features

RGBD Saliency Net

This is the source code of our paper "Learning RGB-D Salient Object Detection using background enclosure, depth contrast, and top-down features".

Our code is implemented based on ELDNet which is RGB saliency detection system. We also make use of gSLICr in our system.

Usage

• Supported OS: We tested our code on Ubuntu 14.04.

• Dependencies: Basically see Caffe installation. We tested our code on CUDA 8.0, OpenCV 3.0.0.

• Installation

  1. We added scripts to original caffe. Please build our version caffe using CMake:

  # execute these command at the root of this directory
  cd caffe && mkdir build && cd build
  cmake ..
  make -j8

  1. Adjust library paths in CMakeList.txt and build code for test.

  # execute these command at the root of this directory
  edit CMakeList.txt
  mkdir build && cd build
  cmake ..
  make

• Run demo program

  sh demo.sh

If you want to test NJUDS2000 dataset images, please use NJUDS2000.caffemodel.

How to create fill and gap maps

  # execute these command at the root of this directory
  cd create_fill_gap && mkdir build && cd build
  cmake ..
  make
  # PLEASE EDIT create_fill_gap.sh TO FIT YOUR ENVIRONMENT
  sh create_fill_gap.sh

Results in our paper

All saliency map outputs are contained in a paper_results.zip file.

Citing our work

Please kindly cite our work if it helps your research:

@InProceedings{Shigematsu_2017_ICCV,
author = {Shigematsu, Riku and Feng, David and You, Shaodi and Barnes, Nick},
title = {Learning RGB-D Salient Object Detection Using Background Enclosure, Depth Contrast, and Top-Down Features},
booktitle = {The IEEE International Conference on Computer Vision (ICCV) Workshops},
month = {Oct},
year = {2017}
}