search

langmanbusi / Semantic-Aware-Low-Light-Image-Enhancement

Semantic-Aware LLIE. CVPR 2023
MIT License
189 stars 7 forks source link

Fine tune the result #11

Closed FireWallDragonDarkFluid closed 1 year ago

FireWallDragonDarkFluid commented 1 year ago

Hi! From the code here

You fine tune the result by using the test data, am I correct?

Which is a serious problem and unfair way to evaluate your model, will you rerun the code by modify the test.py file? I assume there will be a huge different results compare to the orignals.

langmanbusi commented 1 year ago

We use the same way as LLFlow to conduct the comparison, please refer to https://github.com/wyf0912/LLFlow/blob/f5ad48719285be2bc945ebccf8ad2338cad887f6/code/test.py#L144