See this excerpt from the paper. I’m struggling to understand the statement “each dustbin has as many matches as there are keypoints in the other set” from the SuperGlue paper.
Specifically, I’m confused about the scenario where every keypoint in set A is matched with a keypoint in set B. In this case, the N+1th column of the matrix, which represents the dustbin for unmatched keypoints, would be all zeros. How can the sum of this column then equal M, as the paper seems to suggest? What am I missing here?
New contributor
kahan is a new contributor to this site. Take care in asking for clarification, commenting, and answering.
Check out our Code of Conduct.