search

alterzero / DBPN-Pytorch

The project is an official implement of our CVPR2018 paper "Deep Back-Projection Networks for Super-Resolution" (Winner of NTIRE2018 and PIRM2018)
https://alterzero.github.io/projects/DBPN.html
MIT License
565 stars 164 forks source link

The question about train. #46

Open lcj951 opened 5 years ago

lcj951 commented 5 years ago

Hello! Recently,i read the paper of DBPN,i think it's a perfect job. Then i try to run your code to achieve your result at the paper but still a little gap approximately 1.5dB. So i have a question which hope you can teach me something about it . I want to know which the train dataset you use for this work : DIV2K? Flickr? ImageNet dataset? or use all of them ? Best wishes to you! Thankyou!

lchunleo commented 5 years ago

Is it possible to fine tune based on the given pretrained weights for the images that I have? May I check how to do so?Thank you

lcj951 commented 5 years ago

Thank you for answering my question. I train the network use the model of DBPN-LL , the dataset i use to train for my net is DIV2K and Flickr30K , i have test your model with the Set5 dataset but the result is still a gap with the paper's result. So i want to know which dataset you use to train your net to achieve your paper's result(PSNR) ? Hope you can talk something about it. Thank you very much . Best wishes to you! Thankyou!

1136831151 commented 4 years ago

how gpus cost