Continuous Feature Representation for Camouflaged Object Detection

Authors: Ze Song, Xudong Kang, Xiaohui Wei, Jinyang Liu, Zheng Lin, and Shutao Li.

Code implementation of "Continuous Feature Representation for Camouflaged Object Detection". IEEE TIP 2025.Paper

Prerequisites

Install Prerequisites with the following command:

conda create -n CFRNet python = 3.7
conda install pytorch==1.6.0 torchvision==0.7.0 cudatoolkit=10.2 -c pytorch

Usage

1. Download pre-trained Swin transformer model

Please download model from the official websites:

• github or baidu with the fetch code: swin.
• move it into ./pretrained_ckpt/

2. Prepare data

We use data from publicly available datasets:

• downloading testing dataset and move it into ./Dataset/TestDataset/, which can be found in Google Drive or Baidu Drive (extraction code: fapn).

• downloading training/validation dataset and move it into ./Dataset/TrainValDataset/, which can be found in Google Drive or Baidu Drive (extraction code: fapn).

3. Train

To train CFRNet with costumed path:

python MyTrain_Val.py --save_path './snapshot/CFRNet/'

4. Test

To test with trained model:

python MyTesting.py --pth_path './snapshot/CRFNet/Net_epoch.pth'

You can also download prediction maps from Google Drive.

4. Evaluation

We use public one-key evaluation, which is written in MATLAB code (link). Please follow this the instructions in ./eval/main.m and just run it to generate the evaluation results in ./res/.

Citation

Please cite our paper if you find the work useful, thanks!

@article{song2025continuous,
	title={Continuous Feature Representation for Camouflaged Object Detection},
	author={Song, Ze and Kang, Xudong and Wei, Xiaohui and Liu, Jinyang and Lin, Zheng and Li, Shutao},
	journal={IEEE Transactions on Image Processing},
	year={2025},
	publisher={IEEE}
}

⬆ back to top