Closed scott-vsi closed 7 months ago
No, not explicitly.
If youre referring to the recent paper, it would be interesting to modify their approach and use roma matches instead of loftr.
I think dense matching is probably quite good for doppelganger detection.
Ok. Thank you! This and DKM are great repos!
I was wondering if you had any suggestions on how to deal with doplegangers? Is RoMa trained at all to try to differentiate them?