search

huiyang865 / plugnet

67 stars 14 forks source link

Maybe it is a mistake. #1

Closed lartpang closed 3 years ago

lartpang commented 3 years ago

Great work! Thanks for your contribution.!

Here,

https://github.com/huiyang865/plugnet/blob/d4c3c5f150ee61fb1b9cf1d9bc38ee8081ae3022/lib/models/rcan.py#L103

Maybe it should be like this: x = self.tail(res)

huiyang865 commented 3 years ago

Yes, it has been fixed. Thanks.

minglii1998 commented 3 years ago

Hello there, sorry to bother ! @lartpang Can you reproduce the same accuracy as proposed in the paper? My result has a really big gap compared with proposed result and actually I didn't change any configuration (except for some paths) in main_train.sh. It really upsets me.

lartpang commented 3 years ago

@minglii1998 I'm sorry I can't help you. I didn't try to run this code directly. I only read the code to refer to the idea of super resolution. Regarding the question of accuracy, I suggest you contact the author @huiyang865 directly.