Home

Awesome

DSNet

Dynamic Selective Network for RGB-D Salient Object Detection

This repo is an official implementation of the DSNet, which has been accepted in the journal IEEE Transactions on Image Processing, 2021.

DSNet

vis

Prerequisites

Usage

1. Clone the repository

git clone https://github.com/Brook-Wen/DSNet.git
cd DSNet/

2. Training

python main.py --gpu '0' --lr 1e-5 --batch_size 4

3. Testing

python main.py --gpu '0' --batch_size 1 --mode='test' --model='[YOUR PATH]' --test_fold='[SAVE PATH]' --sal_mode='[DATASET]'

4. Evaluation

Citation

If you think this work is helpful, please cite

@inproceedings{wen2021dsnet,
  title={Dynamic Selective Network for RGB-D Salient Object Detection},
  author={Wen, Hongfa and Yan, Chenggang and Zhou, Xiaofei and Cong, Runmin and Sun, Yaoqi and Zheng, Bolun and Zhang, Jiyong and Bao, Yongjun and Ding, Guiguang},
  booktitle={IEEE Transactions on Image Processing},
  year={2021}
}