some question about the code

[AndyYuan96]

Hi，thanks for sharing the code，after reading the code，I wonder why you add dcn and stage_with_dcn to image backbone config, and use mask-rcnn trained in nuimage which config don't have dcn, and why you add other fpn layer to process_detector_feat, as I see in paper figure 2, you don't plot the fpn layer.

[wangzt-halo]

The normal convolution can replaced with the deformable convolution seamlessly thanks to the 'zero init' property of the deformable offsets. There is actually 1 fpn layer in process_detector_feat, which works as a convolutional layer.