This repo. is an official implementation of the BgNet , which has been accepted in the journal Knowledge-based systems, 2022.
The main pipeline is shown as the following,
And some results are presented
>= Pytorch 1.0.0
OpenCV-Python
[optional] matlab
Training and testing datasets are available at
(Baidu) [code:q5e4]
(Google)
python train.py
python test.py
We provide the trained model file (Baidu Res2Net) [code:1fau] (Baidu ResNet) [code:lzzo]
(Google Res2Net) (Google ResNet)
The prediction maps are also available (Res2Net)[code:rbxb] (ResNet)[code:uajy]
Please cite the BgNet in your publications if it helps your research:
@article{CHEN2022,
title = {Boundary-guided Network for Camouflaged Object Detection},
author = {Tianyou Chen and Jin Xiao and Xiaoguang Hu and Guofeng Zhang and Shaojie Wang},
journal = {Knowledge-Based Systems},
year = {2022},
}