Alligat0R: Pre-Training Through Co-Visibility Segmentation for Relative Camera Pose Regression
Fuente:
arXiv
Salvato in:
| Autori principali: | , , |
|---|---|
| Natura: | Preprint |
| Pubblicazione: |
2025
|
| Soggetti: | |
| Accesso online: | |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866908685310099456 |
|---|---|
| author | Loiseau, Thibaut Bourmaud, Guillaume Lepetit, Vincent |
| author_facet | Loiseau, Thibaut Bourmaud, Guillaume Lepetit, Vincent |
| contents | Pre-training techniques have greatly advanced computer vision, with CroCo's cross-view completion approach yielding impressive results in tasks like 3D reconstruction and pose regression. However, cross-view completion is ill-posed in non-covisible regions, limiting its effectiveness. We introduce Alligat0R, a novel pre-training approach that replaces cross-view learning with a covisibility segmentation task. Our method predicts whether each pixel in one image is covisible in the second image, occluded, or outside the field of view, making the pre-training effective in both covisible and non-covisible regions, and provides interpretable predictions. To support this, we present Cub3, a large-scale dataset with 5M image pairs and dense covisibility annotations derived from the nuScenes and ScanNet datasets. Cub3 includes diverse scenarios with varying degrees of overlap. The experiments show that our novel pre-training method Alligat0R significantly outperforms CroCo in relative pose regression. Code is available at https://github.com/thibautloiseau/alligat0r. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2503_07561 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Alligat0R: Pre-Training Through Co-Visibility Segmentation for Relative Camera Pose Regression Loiseau, Thibaut Bourmaud, Guillaume Lepetit, Vincent Computer Vision and Pattern Recognition Pre-training techniques have greatly advanced computer vision, with CroCo's cross-view completion approach yielding impressive results in tasks like 3D reconstruction and pose regression. However, cross-view completion is ill-posed in non-covisible regions, limiting its effectiveness. We introduce Alligat0R, a novel pre-training approach that replaces cross-view learning with a covisibility segmentation task. Our method predicts whether each pixel in one image is covisible in the second image, occluded, or outside the field of view, making the pre-training effective in both covisible and non-covisible regions, and provides interpretable predictions. To support this, we present Cub3, a large-scale dataset with 5M image pairs and dense covisibility annotations derived from the nuScenes and ScanNet datasets. Cub3 includes diverse scenarios with varying degrees of overlap. The experiments show that our novel pre-training method Alligat0R significantly outperforms CroCo in relative pose regression. Code is available at https://github.com/thibautloiseau/alligat0r. |
| title | Alligat0R: Pre-Training Through Co-Visibility Segmentation for Relative Camera Pose Regression |
| topic | Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2503.07561 |