You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Hello author, I would like to use online data augmentation in BCNet now. I see that the image size can be randomly processed in the cfg file,So I want to add some other enhancements, such as the ones shown below,
T.RandomBrightness(0.9, 1.1) and T.RandomFlip(prob=0.5)
and I'm not sure how to add them. Should I modify the datamapper.py file? Or should I add some code to my train.py file? I'm looking forward to your reply and i will be grateful if u can help me! Please forgive my poor English.
Hello author, I would like to use online data augmentation in BCNet now. I see that the image size can be randomly processed in the cfg file,So I want to add some other enhancements, such as the ones shown below,
T.RandomBrightness(0.9, 1.1) and T.RandomFlip(prob=0.5)
and I'm not sure how to add them. Should I modify the
datamapper.py
file? Or should I add some code to mytrain.py
file? I'm looking forward to your reply and i will be grateful if u can help me! Please forgive my poor English.These are some related works that I found, and i hope these can help us.
https://stackoverflow.com/questions/71774744/how-to-use-detectron2s-augmentation-with-datasets-loaded-using-register-coco-in/71799464#71799464
https://github.com/facebookresearch/detectron2/issues/1578
The text was updated successfully, but these errors were encountered: