[IJCV-2024] ViDSOD-100: A New Dataset and a Baseline Model for RGB-D Video Salient Object Detection

This repository is the official implementation of ViDSOD-100

Dataset

You can download our ViDSOD-100 dataset from here.
For both training and inference, the following dataset structure is required:

vidsod_100
|-- train
|-- train_flow
|-- test
|-- test_flow

training

sh train.sh

inference

sh infer.sh

evaluation

sh cal_scores.sh

Weights

We released our checkpoint on ViDSOD-100

Saliency Maps

You can download the saliency maps generated by our method as well as by SOTA methods at the following link.

Citation

If you find our work useful for your research, please cite us:

@article{lin2024vidsod,
  title={ViDSOD-100: A New Dataset and a Baseline Model for RGB-D Video Salient Object Detection},
  author={Lin, Junhao and Zhu, Lei and Shen, Jiaxing and Fu, Huazhu and Zhang, Qing and Wang, Liansheng},
  journal={International Journal of Computer Vision},
  pages={1--19},
  year={2024},
  publisher={Springer}
}