Home

Awesome

JL-DCF (CVPR 2020, TPAMI 2021)

JL-DCF: Joint Learning and Densely-Cooperative Fusion Framework for RGB-D Salient Object Detection (CVPR2020) [PDF][中文版]
Siamese Network for RGB-D Salient Object Detection and Beyond (TPAMI2021) [PDF][中文版]
-Testing code is released!
:fire:News!!!:fire: The Pytorch implementation (Training & Testing) of JL-DCF is now released! Please step to the new Project Page for more information.
-Notice that there was a small description mistake in our papers that the NJU2K dataset we used indeed has 1,985 samples (with which we actually experimented), rather than 2,000.

alt text

alt text

alt text

Usage

The original implementation of JL-DCF is Caffe. The code can run on both Windows and Linux, depending on which OS you are using. It has been tested on Linux (with Ubuntu 16.04, CUDA-10, Cudnn-7.6, Matlab 2015b) and also Windows 10 (with CUDA-9, Cudnn-7.6, Matlab 2018a, Visual Studio 2015). It should also work on other configurations (better on CUDA-8.0+, Cudnn-5.0+, and Matlab 2013a+) but we didn't try.

  1. STEP1: To run the code, you should first install Caffe and also its MATLAB Interface.

    Suggested Caffe version from Github
    For Linux: https://github.com/BVLC/caffe
    For Windows: https://github.com/BVLC/caffe/tree/windows OR https://github.com/happynear/caffe-windows (the latter works more easily with Visual Studio project *.sln and is recommended to run our code for Windows users)

    Note that standard Caffe is enough for running our code!

  2. STEP2: Download the model of JL-DCF from the following links and have it in the "models" folder.
    链接：https://pan.baidu.com/s/1hHckF5PPtFXM52GaYNeuOw 提取码：nvoi
    or:
    https://drive.google.com/open?id=185L3uULu0-GKWyLxE6S8X79pYYj6JUOg

  3. STEP3: Open MATLAB, run demo_JLDCF.m and get the results in "results" folder.

Results

Results of our JL-DCF model on 7 benchmark datasets (NJU2K, NLPR, STERE, RGBD135, LFSD, SIP, DUT-RGBD) can be found below:
链接: https://pan.baidu.com/s/1NGaoZbmPKDr1auMtO0syFQ 提取码: osnf
or:
https://drive.google.com/open?id=1I_2i5XbjTdAVfgOh4wpfV3EuqVaXRQH0

References of datasets:
[STERE] Leveraging stereopsis for saliency analysis. In CVPR 2012.
[NJU2K] Depth saliency based on anisotropic centersurround difference. In ICIP 2014.
[NLPR] Rgbd salient object detection: A benchmark and algorithms. In ECCV 2014.
[RGBD135] Depth enhanced saliency detection method. In International Conference on Internet Multimedia Computing and Service. ACM, 2014.
[LFSD] Saliency detection on light field. In CVPR 2014.
[DUT-RGBD] Depth-induced multi-scale recurrent attention network for saliency detection. In ICCV 2019.
[SIP] Rethinking RGB-D salient object detection: Models, datasets, and large-scale benchmarks. IEEE TNNLS, 2020.

:fire:As some researchers require, results of our JL-DCF model on SSD dataset (having 80 samples) can be found below:
链接: https://pan.baidu.com/s/18q3YNCfEv9o5uwMUeJ3rgw 提取码: drir
or:
https://drive.google.com/open?id=1Yqc7rQotSFjvqcnDoOavPTur81U4v0vj

References of dataset:
[SSD] A three-pathway psychobiological framework of salient object detection using stereoscopic technology. In ICCVW 2017.

Important Tips!

Note that our JL-DCF model was trained on depth maps which satisfy the rule that closer objects present lower depth values (are "black"), while further objects have higher depth values (are "white"). such a rule is enforced in order to meet physical common sense. We observed that the model performance would somewhat degrade when using reversed maps (e.g., disparity maps) during testing. <strong>So be aware of the following issues when testing the models: </strong>

  1. Depth maps are min-max normalized into [0, 1] or [0, 255].
  2. Closer objects present lower depth values (are "black"). alt text

Dataset

Our training dataset is
链接: https://pan.baidu.com/s/1uF6LxbH0RIcMFN71cEcGHQ 提取码: 5z48
or:
https://drive.google.com/open?id=1BpVabSlPH_GhozzRQYjxTOT_cS6xDUgf

Citation

Please cite our paper if you find the work useful:

@inproceedings{Fu2020JLDCF,
title={JL-DCF: Joint Learning and Densely-Cooperative Fusion Framework for RGB-D Salient Object Detection},
author={Fu, Keren and Fan, Deng-Ping and Ji, Ge-Peng and Zhao, Qijun},
booktitle={IEEE Conference on Computer Vision and Pattern Recognition (CVPR)},
pages={3052--3062},
year={2020}
}
    
@article{Fu2021siamese,
title={Siamese Network for RGB-D Salient Object Detection and Beyond},
author={Fu, Keren and Fan, Deng-Ping and Ji, Ge-Peng and Zhao, Qijun and Shen, Jianbing and Zhu, Ce},
journal={IEEE Transactions on Pattern Analysis and Machine Intelligence (TPAMI)},
year={2021}
}

Benchmark RGB-D SOD

The complete RGB-D SOD benchmark can be found in this page
http://dpfan.net/d3netbenchmark/