Tao Huang, Songjiang Li, Xu Jia, Huchuan Lu, Jianzhuang Liu

Abstract: In the last few years, image denoising has benefited a lot from the fast development of neural networks. However, the requirement of large amounts of noisy-clean image pairs for supervision limits the wide use of these models. Although there have been a few attempts in training an image denoising model with only single noisy images, existing self-supervised denoising approaches suffer from inefficient network training, loss of useful information, or dependence on noise modeling. In this paper, we present a very simple yet effective method named Neighbor2Neighbor to train an effective image denoising model with only noisy images. Firstly, a random neighbor sub-sampler is proposed for the generation of training image pairs. In detail, input and target used to train a network are images sub-sampled from the same noisy image, satisfying the requirement that paired pixels of paired images are neighbors and have very similar appearance with each other. Secondly, a denoising network is trained on sub-sampled training pairs generated in the first stage, with a proposed regularizer as additional loss for better performance. The proposed Neighbor2Neighbor framework is able to enjoy the progress of state-of-the-art supervised denoising networks in network architecture design. Moreover, it avoids heavy dependence on the assumption of the noise distribution. We explain our approach from a theoretical perspective and further validate it through extensive experiments, including synthetic experiments with different noise distributions in sRGB space and real-world experiments on a denoising benchmark dataset in raw-RGB space.

Official Pytorch implementation for the paper accepted by CVPR 2021.

• Arxiv
• Conference
• Supplementary

This code was tested on:

• Python 3.7
• Pytorch 1.3

Images in the training set are from the ImageNet validation set with size between 256x256 and 512x512 pixels. There are 44328 images in total.

python dataset_tool.py 
--input_dir=./ILSVRC2012_img_val 
--save_dir=./Imagenet_val

• optional arguments:
  • input_dir Path to the ImageNet validation set
  • save_dir Path to save the training set

To train a network, run:

python train.py 
--data_dir=./Imagenet_val 
--val_dirs=./validation 
--noisetype=gauss25 
--save_model_path=./results 
--log_name=unet_gauss25_b4e100r02 
--increase_ratio=2

• selected optional arguments:
  • data_dir Path to the training set
  • val_dirs Path to the validation sets
  • noisetype Distribution of image noise, choosing from gauss25, gauss5_50, poisson30, or poisson5_50
  • save_model_path Base-path to the saved files
  • log_name Path to the saved files
  • increase_ratio Weight for the trade off between the reconstruction term and the regularization term in the loss function

@InProceedings{Huang_2021_CVPR,
    author    = {Huang, Tao and Li, Songjiang and Jia, Xu and Lu, Huchuan and Liu, Jianzhuang},
    title     = {Neighbor2Neighbor: Self-Supervised Denoising From Single Noisy Images},
    booktitle = {Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)},
    month     = {June},
    year      = {2021},
    pages     = {14781-14790}
}