AI Paper Atlas

Research map / Label-efficient and robust learning / Recognition and long-tail

AnyUp: Universal Feature Upsampling

Thomas Wimmer, Prune Truong, Marie-Julie Rakotosaona, Michael Oechsle, Federico Tombari, Bernt Schiele, Jan Eric Lenssen

ICLR 2026 · Label-efficient and robust learning › Recognition and long-tail · approach: Correspondence and causal representation learning · 23 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
add local window attention (ablation)AnyUp · NYUv2 absolute depthRMSE: 0.4854 -> 0.4755Improved
add local window attention (ablation)AnyUp · COCO-StuffmIoU: 62.12 -> 62.16No clear effect
use own data sampling instead of JAFAR sampling (ablation)AnyUp · COCO-StuffmIoU: 62.03 -> 62.16Improved
add self-consistency loss (ablation)AnyUp · COCO-StuffmIoU: 62.09 -> 62.16Improved
add regularization (self-consistency and augmentations) (ablation)AnyUp · COCO-StuffmIoU: 61.90 -> 62.16Improved
add feature path into key computation (ablation)AnyUp · COCO-StuffmIoU: 61.97 -> 62.16Improved
increase filter basis size from 0 to 128 (ablation)AnyUp · COCO-StuffmIoU: 61.97 -> 62.16Improved
increase filter basis size from 128 to 256 (ablation)AnyUp · COCO-StuffmIoU: 62.16 -> 62.19No clear effect
train on multiple backbones instead of DINOv2 only (ablation)AnyUp · DeiT ViT-B (unseen) mIoUmIoU: 53.93 -> 54.83Improved
train on multiple backbones instead of DINOv2 only (ablation)AnyUp · DINOv2 seen features COCOmIoU: 62.16 -> 62.04Hurt
train on test encoder instead of DINOv2 (ablation)AnyUp · SigLIP 2 (LoftUp setup)mIoU: 51.68 -> 54.45Improved
train on DINOv2 instead of DINOv3 for DINOv3 test (ablation)AnyUp · DINOv3 ViT-S+mIoU: 62.96 vs 62.99No clear effect

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy