Salvato in:
Dettagli Bibliografici
Autori principali: Pariza, Valentinos, Salehi, Mohammadreza, Burghouts, Gertjan, Locatello, Francesco, Asano, Yuki M.
Natura: Preprint
Pubblicazione: 2024
Soggetti:
Accesso online:https://arxiv.org/abs/2408.11054
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866917987255058432
author Pariza, Valentinos
Salehi, Mohammadreza
Burghouts, Gertjan
Locatello, Francesco
Asano, Yuki M.
author_facet Pariza, Valentinos
Salehi, Mohammadreza
Burghouts, Gertjan
Locatello, Francesco
Asano, Yuki M.
contents We introduce NeCo: Patch Neighbor Consistency, a novel self-supervised training loss that enforces patch-level nearest neighbor consistency across a student and teacher model. Compared to contrastive approaches that only yield binary learning signals, i.e., 'attract' and 'repel', this approach benefits from the more fine-grained learning signal of sorting spatially dense features relative to reference patches. Our method leverages differentiable sorting applied on top of pretrained representations, such as DINOv2-registers to bootstrap the learning signal and further improve upon them. This dense post-pretraining leads to superior performance across various models and datasets, despite requiring only 19 hours on a single GPU. This method generates high-quality dense feature encoders and establishes several new state-of-the-art results such as +5.5% and +6% for non-parametric in-context semantic segmentation on ADE20k and Pascal VOC, +7.2% and +5.7% for linear segmentation evaluations on COCO-Things and -Stuff and improvements in the 3D understanding of multi-view consistency on SPair-71k, by more than 1.5%.
format Preprint
id arxiv_https___arxiv_org_abs_2408_11054
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Near, far: Patch-ordering enhances vision foundation models' scene understanding
Pariza, Valentinos
Salehi, Mohammadreza
Burghouts, Gertjan
Locatello, Francesco
Asano, Yuki M.
Computer Vision and Pattern Recognition
Artificial Intelligence
We introduce NeCo: Patch Neighbor Consistency, a novel self-supervised training loss that enforces patch-level nearest neighbor consistency across a student and teacher model. Compared to contrastive approaches that only yield binary learning signals, i.e., 'attract' and 'repel', this approach benefits from the more fine-grained learning signal of sorting spatially dense features relative to reference patches. Our method leverages differentiable sorting applied on top of pretrained representations, such as DINOv2-registers to bootstrap the learning signal and further improve upon them. This dense post-pretraining leads to superior performance across various models and datasets, despite requiring only 19 hours on a single GPU. This method generates high-quality dense feature encoders and establishes several new state-of-the-art results such as +5.5% and +6% for non-parametric in-context semantic segmentation on ADE20k and Pascal VOC, +7.2% and +5.7% for linear segmentation evaluations on COCO-Things and -Stuff and improvements in the 3D understanding of multi-view consistency on SPair-71k, by more than 1.5%.
title Near, far: Patch-ordering enhances vision foundation models' scene understanding
topic Computer Vision and Pattern Recognition
Artificial Intelligence
url https://arxiv.org/abs/2408.11054