chaoren88 / CasaPuNet

CasaPuNet: Channel Affine Self-Attention Based Progressively Updated Network for Real Image Denoising (IEEE TII)

CasaPuNet: Channel Affine Self-Attention Based Progressively Updated Network for Real Image Denoising

Environment

The model is built in PyTorch 1.7.1 and tested on Ubuntu 20.04 environment (Python3.7, CUDA10.2).

Download

Pretrained Model: https://drive.google.com/file/d/1lTojt_U10Lj6IzgvrlXRk6p7gMkHDEIQ/view?usp=share_link

DND Dataset: https://noise.visinf.tu-darmstadt.de/downloads/

SIDD Dataset: https://www.eecs.yorku.ca/~kamel/sidd/benchmark.php

CT Core Image Dataset: https://drive.google.com/file/d/1QWPj0OMfbgNT4cgucFjL3YTqmpWNc5c4/view?usp=share_link

Test

Extract the files to dataset folder and checkpoint folder as follow:

~/
  dataset/
    benchmark/
      dnd_2017/
        images_srgb/
            ... (mat files)
            ... (mat files)
        info.mat
      sidd/
        BenchmarkNoisyBlocksSrgb.mat
  checkpoint/
    checkpoint.pth.tar

To test on DND or SIDD Benchmark, run

python test_benchmark.py --type dnd_or_sidd

To test on noisy images, run

python test_image.py