search

yulunzhang / RDN

Torch code for our CVPR 2018 paper "Residual Dense Network for Image Super-Resolution" (Spotlight)
543 stars 108 forks source link

MSE LOSS #9

Open sonal-bansal opened 5 years ago

sonal-bansal commented 5 years ago

Hi, How much loss(MSE) did u obtained after your training got completed, I've trained my model for about 100 epochs and its ~100 ? Moreover, I'm using RandomCrop of patches of LR and HR for training ...how will it effect my training?

Thank you

yulunzhang commented 5 years ago

Hi, I cannot remember the MSE loss exactly. Some of the log files are deleted. Anyway, if you use the training code to train a model with MSE, the validation and test performance should be similar or slightly lower than that using L1 loss.

For the usage of RandomCrop, I think it performs very similar as the original code dose. Thanks.