Is there a way to define validation images separately and evaluate the final results on them? I have checked the codes but I couldn't find any flag or place where I can specify my validation images without randomly selecting them. I would appreciate it if such a setting exists and could help me to do that. Thank you.
Yes this is possible. If you look at the dataloder, you can specify the held-out set logic. Currently we simply use a subset of training set as the evaluation. I hope this helps.
Is there a way to define validation images separately and evaluate the final results on them? I have checked the codes but I couldn't find any flag or place where I can specify my validation images without randomly selecting them. I would appreciate it if such a setting exists and could help me to do that. Thank you.