The author measured grid-induced error across all 499,188 annotated keypoints in SPair-71k.
Continuous feature decoding improves fine-grained semantic matching between images
ImCorr replaces grid-bound feature lookup with a coordinate-conditioned field, improving keypoint matching at sub-patch scales.
Academic
Yusung Choi
Pukyong National University
Research Digest··3 min read
Choi examines why semantic correspondence systems, which locate matching semantic points across images, lose accuracy under strict spatial tolerances.
Why this paper
From Pukyong National University
In one line
ImCorr eliminates grid quantization errors in semantic correspondence via continuous feature decoding.
What we could check
- ·No code link found
- ·No weights link found
- ·No dataset link found
- ·No compute details found
- ·No stated limitations found
- ✓Reports numbers on named benchmarks
Observed from the paper text and links we have. Absence here means we did not find it, not that it does not exist.
§