search

chaofengc / PSFRGAN

PyTorch codes for "Progressive Semantic-Aware Style Transformation for Blind Face Restoration", CVPR2021
Other
370 stars 68 forks source link

try to train 1024 resolution memory out #34

Closed goodclass closed 3 years ago

goodclass commented 3 years ago

Hi, this is a great code

I try to train 1024 sizes， Gout_size=1024 and batch_size=1 use RTX 3090 24G，Train one memory out of Can some configurations be modified to adapt to a sample training

chaofengc commented 3 years ago

This is likely because of the style loss and 19 semantic maps. You may try to reduce the region number of parsing maps. It is not that necessary to use all 19 regions.