BCDNet

Keunsoo Ko, Yeong Jun Koh, and Chang-Su Kim

Official PyTorch Code for "Blind and Compact Denoising Network Based on Noise Order Learning, IEEE Trans. Image Process., 2022"

Installation

Download repository:

    $ git clone https://github.com/keunsoo-ko/BCDNet.git

Download pre-trained model parameters

Usage

Run Test for real noise dataset on the SSID dataset:

    $ python demo.py --data_path "Path" --model_path BCDNet_Real.pth(put downloaded model path)

The path of dataset should be "Path", in which consisting of "Clean" and "Noise" folders as below

├─ Path
    ├── Clean
    └── Noise

If you get the performance in paper, please use the self-ensemble stratey as mentioned in the paper.