README.md

Image restoration using the Pix2Pix [1] cGan and U-net [2] networks

Pix2Pix network is used for image inpaiting. Afterwards, back-to-back U-net networks are used for denoising and deblurring effect.

References:

[1] https://arxiv.org/pdf/1611.07004.pdf

[2] https://arxiv.org/pdf/1505.04597.pdf

Examples:

            (a) noisy_image                 (b) inpaiting_result_for(a)           (c) denoise_deblur_image             (d) ground_truth

            (a) noisy_image                 (b) inpaiting_result_for(a)           (c) denoise_deblur_image             (d) ground_truth

            (a) noisy_image                 (b) inpaiting_result_for(a)            (c) denoise_deblur_image

            (a) noisy_image                 (b) inpaiting_result_for(a)            (c) denoise_deblur_image

Requirements

Python (suggested 3.7.1)  
Numpy   
os-sys  
OpenCv  
Tensorflow (suggested 2.4.0)