search

sanghyun-son / EDSR-PyTorch

PyTorch version of the paper 'Enhanced Deep Residual Networks for Single Image Super-Resolution' (CVPRW 2017)
MIT License
2.42k stars 668 forks source link

Can I get your pytorch model on benchmark datasets #32

Closed FLT19940317 closed 6 years ago

FLT19940317 commented 6 years ago

Hello, your work is amazing. Can I get your Pytorch model on benchmark datasets. I want to do some research based on your EDSR.

sanghyun-son commented 6 years ago

Hello.

Do you mean that you want to reproduce the result on benchmark datasets (Set5, Set14, Urban100, B100) from the paper?

We use same models for DIV2K and benchmark evaluations, so you can easily get the results.

Please see these scripts!

If I misunderstood your question, let me know.

Thank you.

FLT19940317 commented 6 years ago

Hello, Thank for your reply. Paper's benchmark evaluations are only on the Y channel. My understanding is first transfer bechemark datasets to the Ycbcr color space, and use only Y channel as input of network resulting in output of one channel. But your trained model's input are 3 channel. So is it means that you evaluate benchmarck datasets using the RGB image as input and transfering RGB output to Ycbcr for evaluating? I don't know if I make myself clear? Thank you very much!

sanghyun-son commented 6 years ago

You understood it correctly.

We used RGB inputs and transformed outputs to YCbCr for evaluation.

FLT19940317 commented 6 years ago

Ok, Thank you very much. Your work is great!