(TCSVT 2024) D3C2-Net: Dual-Domain Deep Convolutional Coding Network for Compressive Sensing

Weiqi Li, Bin Chen, Shuai Liu, Shijie Zhao, Bowen Du, Yongbing Zhang and Jian Zhang

School of Electronic and Computer Engineering, Peking University

Accepted for publication as a Regular paper in the IEEE Transactions on Circuits and Systems for Video Technology (TCSVT).

Core Ideas

Environment

pip install -r requirements.txt

Train

Download the dataset of Waterloo Exploration Database and put all images in the pristine_images directory (containing 4744 .bmp image files) into ./data/train, then run:

CUDA_VISIBLE_DEVICES=0 torchrun --nproc_per_node=1 --master_port=35001 train.py --phase_num 25 --learning_rate 1e-4 --batch_size 8

The log and model files will be in ./log and ./model, respectively.

Test

The model checkpoint file is provided in ./model, and the test sets are in ./data.

python test.py

Supplementary Materials

We provide theorem proof and more applications of D3C2-Net in supplementary materials.

Citation

If you find the code helpful in your research or work, please cite the following paper:

@article{li2024d3c2,
  title={D3C2-Net: Dual-Domain Deep Convolutional Coding Network for Compressive Sensing},
  author={Weiqi, Li and Bin, Chen and Shuai, Liu and Shijie, Zhao and Bowen, Du and Yongbing, Zhang and Jian, Zhang},
  journal={IEEE Transactions on Circuits and Systems for Video Technology},
  year={2024},
  publisher={IEEE}
}