SueeH

根据代码：gen_imglist 只是将所有数据合并到一起，为训练提供了数据的总数(训练代码中读取了总数)。产生tfrecord还是pos/part/neg/landmark 4个文件。不知道理解的对不对

> 时间太久，不是记得很清楚了。

I have trained on EBB dataset with Pytorch PyNET. Details: Training on level2, psnr on valid set(sampled some pic from ebb) can reach ~23. But on level1, psnr on same...

>  Hi I find train dataset(with & without bokeh), but no bokeh image of valid or test sets, could you share the 200 valid bokeh image? Thanks a lot.

Thank you for your reply! I will try different input and follow your works. Best wishes！

I find the data struct is different from described on website. So you read like this: 'P1020171/P1020171.jpg' and 'P1020171/P1020171LinesAndVP.mat'? Also key_word "lines" is in *LinesAndVP.mat.

> > I find the data struct is different from described on website. > > So you read like this: 'P1020171/P1020171.jpg' and 'P1020171/P1020171LinesAndVP.mat'? > > Also key_word "lines" is in...

> > > > I find the data struct is different from described on website. > > > > So you read like this: 'P1020171/P1020171.jpg' and 'P1020171/P1020171LinesAndVP.mat'? > > >...

> @SueeH I am thinking about doing the similar thing. How did you exactly fine-tune the model on Deepglint data set? I can not get your points exactly, could you...

All the params same with authors, I just fine-tune the lr, keep other params the same Since my gpu memory limited, 45 for each is full use of gpu memory....