关于MSDeformAttn中sampling_locations的归一化如何理解问题

fundamentalvision / Deformable-DETR

Deformable DETR: Deformable Transformers for End-to-End Object Detection.

Apache License 2.0

3.22k stars 520 forks source link

Open 20231211 opened 2 months ago

20231211 commented 2 months ago

elif reference_points.shape[-1] == 4: sampling_locations = reference_points[:, :, None, :, None, :2] \

sampling_offsets / self.n_points reference_points[:, :, None, :, None, 2:] 0.5 这里归一化为什么是除以 self.n_points reference_points[:, :, None, :, None, 2:] 0.5