You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Does all image pair in the training data set contain at least one matching feature points?
Should I need to add a completely mismatched image in the training set?
The text was updated successfully, but these errors were encountered:
we just released the source code for training Superglue-like matchers from scratch at https://github.com/feixue94/imp-release. From our experiments, we can reach close even better performance than SuperGlue.
Does all image pair in the training data set contain at least one matching feature points?
Should I need to add a completely mismatched image in the training set?
The text was updated successfully, but these errors were encountered: