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
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 changed | Where | Measured | Effect |
|---|---|---|---|
| add local window attention (ablation) | AnyUp · NYUv2 absolute depth | RMSE: 0.4854 -> 0.4755 | Improved |
| add local window attention (ablation) | AnyUp · COCO-Stuff | mIoU: 62.12 -> 62.16 | No clear effect |
| use own data sampling instead of JAFAR sampling (ablation) | AnyUp · COCO-Stuff | mIoU: 62.03 -> 62.16 | Improved |
| add self-consistency loss (ablation) | AnyUp · COCO-Stuff | mIoU: 62.09 -> 62.16 | Improved |
| add regularization (self-consistency and augmentations) (ablation) | AnyUp · COCO-Stuff | mIoU: 61.90 -> 62.16 | Improved |
| add feature path into key computation (ablation) | AnyUp · COCO-Stuff | mIoU: 61.97 -> 62.16 | Improved |
| increase filter basis size from 0 to 128 (ablation) | AnyUp · COCO-Stuff | mIoU: 61.97 -> 62.16 | Improved |
| increase filter basis size from 128 to 256 (ablation) | AnyUp · COCO-Stuff | mIoU: 62.16 -> 62.19 | No clear effect |
| train on multiple backbones instead of DINOv2 only (ablation) | AnyUp · DeiT ViT-B (unseen) mIoU | mIoU: 53.93 -> 54.83 | Improved |
| train on multiple backbones instead of DINOv2 only (ablation) | AnyUp · DINOv2 seen features COCO | mIoU: 62.16 -> 62.04 | Hurt |
| train on test encoder instead of DINOv2 (ablation) | AnyUp · SigLIP 2 (LoftUp setup) | mIoU: 51.68 -> 54.45 | Improved |
| train on DINOv2 instead of DINOv3 for DINOv3 test (ablation) | AnyUp · DINOv3 ViT-S+ | mIoU: 62.96 vs 62.99 | No clear effect |
Closest papers by meaning
- FeatUp: A Model-Agnostic Framework for Features at Any Resolution (ICLR 2024 · similarity 0.86)
- JAFAR: Jack up Any Feature at Any Resolution (NeurIPS 2025 · similarity 0.84)
- TinySAM: Pushing the Envelope for Efficient Segment Anything Model (AAAI 2025 · similarity 0.82)
- Stable Segment Anything Model (ICLR 2025 · similarity 0.82)
- Overcoming the Pitfalls of Vision-Language Model Finetuning for OOD Generalization (ICLR 2024 · similarity 0.82)
- Depth Anything: Unleashing the Power of Large-Scale Unlabeled Data (CVPR 2024 · similarity 0.82)
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.
