Did you try training them system using noisy images, like what is done in papers like Noise2Noise, Noise2void and Noise2self?

Owlz / Face-Denoising

Group project for master class "Context Aware Security Analysis for Computer Vision"

6 stars 4 forks source link

Did you try training them system using noisy images, like what is done in papers like Noise2Noise, Noise2void and Noise2self? #6

Open abhinavclemson opened 4 years ago

Owlz commented 4 years ago

We did train this models with noisy pictures as inputs and non-noisy pictures as expected output. We had to use our own dataset of noisy pictures because the focus was on human faces