Awesome

ModalityProfile_MM23

The official implementation of ACM MM 2023 “Modality Profile - A New Critical Aspect to be Considered When Generating RGB-D Salient Object Detection Training Set”.

:fire: You are now able to download the proposed dataset and the corresponding results!

:loudspeaker:Download the proposed dataset. In this paper, we constructed a large training set containing 40k RGBD images, and all images are automatically equipped with high-quality pseudo-GTs. The details can be found in our paper.

:loudspeaker:Download our results (saliency maps). We retrained the BBS, SSL, and SPSN models on our proposed dataset (MP2K and MP4K), resulting in significant performance enhancements. The test sets include COME-E, COME-H, DES, DUT, LFSD, NJU2K, NLPR, ReDWeb, SIP, SSD and STERE.

All 11 test sets can be downloaded from this page, and the evaluation code can be found here. Please note that both the test sets and evaluation code are compiled from existing works, and we acknowledge the contributions of these resources.

:hearts:Should our work prove beneficial to your research, we would appreciate citation of our paper.

@inproceedings{wang2023modality,
  title={Modality Profile-A New Critical Aspect to be Considered When Generating RGB-D Salient Object Detection Training Set},
  author={Wang, Xuehao and Li, Shuai and Chen, Chenglizhao and Hao, Aimin and Qin, Hong},
  booktitle={Proceedings of the 31st ACM International Conference on Multimedia},
  pages={3355--3364},
  year={2023}
}

Meanwhile, our previous works have also emphasized the importance of modality quality. We recommend that you consider reading these papers, as they have significantly contributed to the advancement of the RGBD SOD community. They have published in TIP and Neurocomputing.

@article{wang2020data,
  title={Data-level recombination and lightweight fusion scheme for RGB-D salient object detection},
  author={Wang, Xuehao and Li, Shuai and Chen, Chenglizhao and Fang, Yuming and Hao, Aimin and Qin, Hong},
  journal={IEEE Transactions on Image Processing},
  volume={30},
  pages={458--471},
  year={2020},
  publisher={IEEE}
}

@article{wang2021depth,
  title={Depth quality-aware selective saliency fusion for RGB-D image salient object detection},
  author={Wang, Xuehao and Li, Shuai and Chen, Chenglizhao and Hao, Aimin and Qin, Hong},
  journal={Neurocomputing},
  volume={432},
  pages={44--56},
  year={2021},
  publisher={Elsevier}
}

:email:<a href="mailto:wangxuehao1993@163.com">Please feel free to contact me</a>