the output channals of FeatureAlign_V2

[ChengYi1996]

self.dcpack_L2 = dcn_v2(out_nc, out_nc, 3, stride=1, padding=1, dilation=1, deformable_groups=8, extra_offset_mask=True) why out_nc is 256, not 216(3 x kernel_size x kernal_size x deformable_groups) ?

[chenhaiwen]

I also have the same question

[JinYangprominent1994]

216 is the dimension of the offset tensor (which is the one of inputs of this DC block), and the dimension of the whole deformable convolution block will be 256