Prediction on full size image without transformation ?

PonteIneptique commented 3 years ago

Hi ! I launched a training session on your model and I have few questions if I may:

Training seems to only output latest_net_G.pth (I get an error FileNotFoundError: [Errno 2] No such file or directory: 'checkpoints/coco_mask/latest_net_GF.pth' if I don't copy your models weight when running FileNotFoundError: [Errno 2] No such file or directory: 'checkpoints/coco_mask/latest_net_GF.pth')
When I finally use your model data, I end up having another issue: it only seems to output 256 (or 512, 768, etc.) px wide and squared image while I'd need to run it on "real world" data. Is there something I need to do for this ?

PonteIneptique commented 3 years ago

Please don't mind the second part of the issue (output on full size), but I'd be happy to have an answer on the first :)

PonteIneptique commented 3 years ago

Sorry for the stupid question, I should definitely take the time to play more with notebooks before asking some...

ericsujw / InstColorization