I have a question to ask you about how to modify the Backbone. When I modify the backbone network, none of the objects in the trained model can be recognized and none of the detection boxes are marked.
The second question is whether you did not restrict his classification loss during training? If you want to restrict the classification loss, do you need to modify losses. py?
Thank you for your contribution to this field!
I have a question to ask you about how to modify the Backbone. When I modify the backbone network, none of the objects in the trained model can be recognized and none of the detection boxes are marked.
The second question is whether you did not restrict his classification loss during training? If you want to restrict the classification loss, do you need to modify losses. py?
If you reply to me, I will be very grateful!