nikolasmarkou / blind_image_denoising Goto Github PK

Filter Response Normalization Layer: Eliminating Batch Dependence in the Training of Deep Neural Networks, 2020

Basic model is resnet_color_laplacian_3x5_non_shared_bn_16x3x3_256x256_skip_input_prune.json

A fantastic repository, thank you.

I'm just getting started with it, but I just thought I'd reach out and ask if you'd accept a pull request that trains for human percetual quality rather than MAE / PSNR in future?

I'm thinking a simple way to achieve a partial solution is to retrain on images in a colourspace like OKLAB where perceptual difference is baked in, and the perceived colour difference formula is trivial, instead of monstrous!

I was also thinking 'edge loss' or extra channels for the H&V image gradients during training with L1 loss, discarded after, could be good.

A rolls royce solution might be adversarial loss, perhaps using a secondary network like Netflix vmaf or something?

If there are any resources related to perceptual quality rather than PSNR, please do point me in the right direction :-)

Use "Image Quality Assessment: From Error Visibility to Structural Similarity, 2004" to enhance the loss function

Add code to support multiple datasets at the same time

Add a configuration and train using a convnext block (instead of resnet) which is SOTA for computer vision tasks
A ConvNet for the 2020s

Can We Gain More from Orthogonality Regularizations in Training Deep CNNs?, 2018

Solving Linear Inverse Problems Using the Prior Implicit in a Denoiser