ShirAmir / dino-vit-features

I wonder if using the previous KMeans instance at this point is intentional. I mean the part_algorithm was trained on normalized_all_fg_sampled_descriptors as opposed to common_part_algorithm, which is trained on normalized_all_common_sampled_descriptors.

dino-vit-features/part_cosegmentation.py

Line 273 in 4b023ec

    
           _, common_part_labels = part_algorithm.index.search(normalized_all_common_descriptors.astype(np.float32), 1)

Hi @mateusz-politycki-wttech, thanks for pointing out the issue! it is indeed a bug.
it has been fixed and updated in the repo.

Due note that we recommend using the three stage version only on small sets.

Use of the previous KMeans instance