search

XPixelGroup / BasicSR

Open Source Image and Video Restoration Toolbox for Super-resolution, Denoise, Deblurring, etc. Currently, it includes EDSR, RCAN, SRResNet, SRGAN, ESRGAN, EDVR, BasicVSR, SwinIR, ECBSR, etc. Also support StyleGAN2, DFDNet.
https://basicsr.readthedocs.io/en/latest/
Apache License 2.0
6.64k stars 1.17k forks source link

Difference between train_sr and train_srresnet #138

Open BeepBump opened 5 years ago

BeepBump commented 5 years ago

Hi, I notice that both train_sr and train_srresnet can use resnet model, what's the difference between these two files? Besides, I understand ResNeXt has groups option, does it perform better than resnet in super-resolution? (train_sr has a choice to change groups, do I simply change it and I can get resnext structure?)

xinntao commented 5 years ago

Yes, train_sr can also achieve the train_srresnet.

ResNeXt can only bring marginal improvement, but it is slow to run on GPU. You can have a try. To use the group in the config file is not a good choice because you do not need to change all the conv to group conv. You need to write a new arch to achieve your goal.