Deep Generative Filter for Motion Deblurring

Removing blur during a shaky camera has been a real problem. In order to solve that problem a de-blurring filter was proposed in this work which uses GAN architecture integrated with global skip connections and dense block architecture. 

The original work was implemented in Lua, and as part of the Image and Video Processing course in my undergraduate,  I picked this paper to be implemented from scratch in python using PyTorch as a framework. The Technical Report and source can be found below, this is also been made public in GitHub 

Technical Report: