Hello,thanks for you best work.When I use prepare.py to generate the image 91 x2 h5 file, the size is only 552M, but the size of the file you gave is 6.88GB. What is the reason?

yjn870 / FSRCNN-pytorch

PyTorch implementation of Accelerating the Super-Resolution Convolutional Neural Network (ECCV 2016)

https://arxiv.org/abs/1608.00367

215 stars 59 forks source link

Hello,thanks for you best work.When I use prepare.py to generate the image 91 x2 h5 file, the size is only 552M, but the size of the file you gave is 6.88GB. What is the reason? #2

Open houqian2180320171 opened 4 years ago

marijavella commented 4 years ago

Hi,

By default the code does not do any data augmentation; reducing the file size drastically. I guess the h5 files provided in the repo include data augmentation. You can get these by setting parser.add_argument('--with-aug', action='store_false') instead of parser.add_argument('--with-aug', action='store_true') in 'prepare.py', you will end up with a larger file.

I hope this helps.

wfxiaobo commented 4 years ago

Hello， There is a problem that makes me confused .How to train model for scale 2 or scale 4 on basic of scale 3