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
Excellent work!
My recent work is about 3D reconstruction of trees. The reconstructed tree models have many ghost images and errors. We consider that trees have similar structures and textures, and there may be mismatched image pairs in feature matching. I would like to ask if you think your Doppelgangers method is suitable for application in our work? Also, does Doppelgangers support Hloc? Is there a specific implementation method? I look forward to your answers, thank you!
The text was updated successfully, but these errors were encountered:
Hello, thanks for your interest in our work. About disambiguating doppelgangers of trees, as our method depends on the LoFTR matches to do classification, it would be good to do a sanity check first to see if the LoFTR model can provide reasonable matches of photos of trees or fail on this out-of-domain scenario.
As for supporting Hloc, while we didn't train a model specified for Hloc, you could use our doppelganger classifier to filter out high-probability doppelganger pairs in the scene graph (matches table of colmap database) in step 2.iii of Hloc's general pipeline.
Excellent work!
My recent work is about 3D reconstruction of trees. The reconstructed tree models have many ghost images and errors. We consider that trees have similar structures and textures, and there may be mismatched image pairs in feature matching. I would like to ask if you think your Doppelgangers method is suitable for application in our work? Also, does Doppelgangers support Hloc? Is there a specific implementation method? I look forward to your answers, thank you!
The text was updated successfully, but these errors were encountered: