Near, far: Patch-ordering enhances vision foundation models' scene understanding Pariza, Valentinos Salehi, Mohammadreza Burghouts, Gertjan Locatello, Francesco ; https://orcid.org/0000-0002-4850-0683 Asano, Yuki M. ddc:000 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 +2.3 % and +4.2% for non-parametric in-context semantic segmentation on ADE20k and Pascal VOC, +1.6% and +4.8% 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%. ICLR 2025 info:eu-repo/semantics/conferenceObject doc-type:conferenceObject text http://purl.org/coar/resource_type/c_5794 https://research-explorer.ista.ac.at/record/20036 https://research-explorer.ista.ac.at/download/20036/20109 Pariza V, Salehi M, Burghouts G, Locatello F, Asano YM. Near, far: Patch-ordering enhances vision foundation models’ scene understanding. In: <i>13th International Conference on Learning Representations</i>. ICLR; 2025:72303-72330. eng info:eu-repo/semantics/altIdentifier/isbn/9798331320850 info:eu-repo/semantics/altIdentifier/arxiv/2408.11054 https://creativecommons.org/licenses/by/4.0/ info:eu-repo/semantics/openAccess