do you use different bounding box regression?

[Zealoe]

because the feature will be changed in the net, so do you have to use different bounding box when detection?

[Zardinality]

I dont quite get it, could you expand on this?

[Zealoe]

when changing the pooling layers or the conv layers, the perception field may be affected, so there may be some effects in the bounding box regression operation. are there any changes in this step?

[Zardinality]

Why the features will be changed and by whom will it be changed? If the offset of final deform_psroi_pool were changed, then the bounding box would no doubt be changed.