Awesome

CMWNet

This project provides the codes and results for 'Cross-Modal Weighting Network for RGB-D Salient Object Detection', ECCV 2020. Springer link or arxiv link Homepage

Our code is implemented based on the Caffe of FlowNet2. You should first install and compile the caffe according to the FlowNet2.

Torch Vision

https://github.com/lartpang/CMWNet.pytorch

Overview and Module details

Overview:

Module details:

Saliency maps and Measure results on 8 Datasets

We provide saliency maps (code: 6f2j) and measure results (code: p0y7) of our CMWNet on 8 datasets (STEREO, NJU2K, LFSD, DES, NLPR, SSD, SIP and additional DUT-RGBD).

You can use the evaluation tool to evaluate the result maps.

The parameter amount of our CMWNet is 85.61M.

The FLOPs of our CMWNet is 322.34G.

Testing

1. test_RGBD.prototxt/ is under models/.
2. Download the trained model (code: z2o4) (RGBD_iter_22500.caffemodel), and put it under models/.
3. The datasets are under datasets/, we provide some testing examples on DES dataset.
4. Run test_matlab/test_CMWNet.m.
5. Saliency maps are under salmaps/DES/.

Related works on RGB-D SOD

(TIP_2021_HAINet) Hierarchical Alternate Interaction Network for RGB-D Salient Object Detection.

(TIP_2020_ICNet) ICNet: Information Conversion Network for RGB-D Based Salient Object Detection.

(Survey) RGB-D Salient Object Detection: A Survey.

Citation

    @inproceedings{Li_2020_CMWNet,
            author = {Li, Gongyang and Liu, Zhi and Ye, Linwei and Wang, Yang and Ling, Haibin},
            title = {Cross-Modal Weighting Network for RGB-D Salient Object Detection},
            journal = {European Conference on Computer Vision (ECCV)},
            year = {2020},
            pages = {665-681},
            month = {Aug.},}

If you encounter any problems with the code, want to report bugs, etc.

Please contact me at lllmiemie@163.com or ligongyang@shu.edu.cn.