Saved in:
Bibliographic Details
Main Authors: Dünkel, Olaf, Wimmer, Thomas, Theobalt, Christian, Rupprecht, Christian, Kortylewski, Adam
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2506.05312
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866909801265496064
author Dünkel, Olaf
Wimmer, Thomas
Theobalt, Christian
Rupprecht, Christian
Kortylewski, Adam
author_facet Dünkel, Olaf
Wimmer, Thomas
Theobalt, Christian
Rupprecht, Christian
Kortylewski, Adam
contents Finding correspondences between semantically similar points across images and object instances is one of the everlasting challenges in computer vision. While large pre-trained vision models have recently been demonstrated as effective priors for semantic matching, they still suffer from ambiguities for symmetric objects or repeated object parts. We propose improving semantic correspondence estimation through 3D-aware pseudo-labeling. Specifically, we train an adapter to refine off-the-shelf features using pseudo-labels obtained via 3D-aware chaining, filtering wrong labels through relaxed cyclic consistency, and 3D spherical prototype mapping constraints. While reducing the need for dataset-specific annotations compared to prior work, we establish a new state-of-the-art on SPair-71k, achieving an absolute gain of over 4% and of over 7% compared to methods with similar supervision requirements. The generality of our proposed approach simplifies the extension of training to other data sources, which we demonstrate in our experiments.
format Preprint
id arxiv_https___arxiv_org_abs_2506_05312
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Do It Yourself: Learning Semantic Correspondence from Pseudo-Labels
Dünkel, Olaf
Wimmer, Thomas
Theobalt, Christian
Rupprecht, Christian
Kortylewski, Adam
Computer Vision and Pattern Recognition
Finding correspondences between semantically similar points across images and object instances is one of the everlasting challenges in computer vision. While large pre-trained vision models have recently been demonstrated as effective priors for semantic matching, they still suffer from ambiguities for symmetric objects or repeated object parts. We propose improving semantic correspondence estimation through 3D-aware pseudo-labeling. Specifically, we train an adapter to refine off-the-shelf features using pseudo-labels obtained via 3D-aware chaining, filtering wrong labels through relaxed cyclic consistency, and 3D spherical prototype mapping constraints. While reducing the need for dataset-specific annotations compared to prior work, we establish a new state-of-the-art on SPair-71k, achieving an absolute gain of over 4% and of over 7% compared to methods with similar supervision requirements. The generality of our proposed approach simplifies the extension of training to other data sources, which we demonstrate in our experiments.
title Do It Yourself: Learning Semantic Correspondence from Pseudo-Labels
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2506.05312