I see the model 02. Predict with pre-trained Faster RCNN models — gluoncv 0.11.0 documentation for object detection with resnet50 backbone. Does this mean that the pre-trained model contains resnet50 + Faster_RCNN architecture? Or does the pre-trained model contains Faster_RCNN architecture only?
If want to cut the model and use the resnet 50 part only from which layer do I need to cut this?
I would appreciate any feedback,