Bug after `fix data leakage`

[zhouzq-thu]

Calling softdetect with normalized_coordinates=False,

rdd/src/RDD.py

Line 89 in cacd40e

keypoints, kptscores, scoredispersitys = self.softdetect(K1, normalized_coordinates=False)

rdd/src/RDD.py

Line 158 in cacd40e

keypoints, kptscores, scoredispersitys = self.softdetect(K1, normalized_coordinates=False)

leads the grid is not normalized in grid_sample.

rdd/src/models/soft_detect.py

Lines 126 to 130 in cacd40e

	if normalized_coordinates:
	keypoints_xy = keypoints_xy / keypoints_xy.new_tensor([w - 1, h - 1]) * 2 - 1 # (w,h) -> (-1~1,-1~1)
	kptscore = torch.nn.functional.grid_sample(scores_map[b_idx].unsqueeze(0), keypoints_xy.view(1, 1, -1, 2),
	mode='bilinear', align_corners=True)[0, 0, 0, :] # CxN

Possible Fix

            # compute result keypoints
            keypoints_xy_nms = torch.stack([indices_kpt % w, indices_kpt // w], dim=1)  # Mx2
            keypoints_xy = keypoints_xy_nms + xy_residual
            keypoints_xy_norm = keypoints_xy / keypoints_xy.new_tensor([w - 1, h - 1]) * 2 - 1  # (w,h) -> (-1~1,-1~1)

            kptscore = torch.nn.functional.grid_sample(scores_map[b_idx].unsqueeze(0), keypoints_xy_norm.view(1, 1, -1, 2),
                                                       mode='bilinear', align_corners=True)[0, 0, 0, :]  # CxN

            keypoints.append(keypoints_xy_norm if normalized_coordinates else keypoints_xy)
            scoredispersitys.append(scoredispersity)
            kptscores.append(kptscore)

[xtcpete]

Hi, thanks for bringing this up. This is intended as the interpolator will do normalization.