About coordinates normalization

[nnop]

Hi, Johan,

I have a question about the coordinates normalization in RoMa.

RoMa/roma/models/matcher.py

Lines 565 to 572 in 36389ef

	def to_normalized_coordinates(self, coords, H_A, W_A, H_B, W_B):
	if isinstance(coords, (list, tuple)):
	kpts_A, kpts_B = coords[0], coords[1]
	else:
	kpts_A, kpts_B = coords[...,:2], coords[...,2:]
	kpts_A = torch.stack((2/W_A * kpts_A[...,0] - 1, 2/H_A * kpts_A[...,1] - 1),axis=-1)
	kpts_B = torch.stack((2/W_B * kpts_B[...,0] - 1, 2/H_B * kpts_B[...,1] - 1),axis=-1)
	return kpts_A, kpts_B

As you're using colmap coordinate fasion, the left- and right-most pixel should be 0.5 and W-0.5.
The current code is converting them to [-1+1/W, 1-1/W].
Shouldn't the conversion be: (2*x-1) / (W-1) - 1?
That would make the left- and right-most pixel exactly [-1, 1].

[Parskatt]

Hi, sorry for the late reply.

We don't want to make the left and right-most pixel exactly [-1,1] because we are (hopefully) using align_corners=False everywhere. Thus we want them to be [-1 + 1/h, 1-1/h].