search

jmiller656 / EDSR-Tensorflow

Tensorflow implementation of Enhanced Deep Residual Networks for Single Image Super-Resolution
MIT License
330 stars 107 forks source link

about training #37

Closed suke27 closed 5 years ago

suke27 commented 5 years ago

Hello, I found you did not do normalization for input data, just do reduce mean op. training can work well even without normalization ?

jmiller656 commented 5 years ago

Empirically, it can!

I can't give any answers as to why this is done though, I'm just following the paper and am not one of the authors. So I can't explain why they don't do normalization other than subtracting the mean.