mlsanityData to 14 Oct 2026

AnyUp: Universal Feature Upsampling

arXiv 2510.12764 ↗First posted 14 Oct 2025cs.CV cs.LGcited by 28

Not on the map. Why?

Abstract (arXiv)

We introduce AnyUp, a method for feature upsampling that can be applied to any vision feature at any resolution, without encoder-specific training. Existing learning-based upsamplers for features like DINO or CLIP need to be re-trained for every feature extractor and thus do not generalize to different feature types at inference time. In this work, we propose an inference-time feature-agnostic upsampling architecture to alleviate this limitation and improve upsampling quality. In our experiments, AnyUp sets a new state of the art for upsampled features, generalizes to different feature types, and preserves feature semantics while being efficient and easy to apply to a wide range of downstream tasks.

Key facts. Cited by 28 papers in the map (data through 14 Oct 2026). Builds on 33 of its 43 references that are in the map. First posted on arXiv 14 Oct 2025.

Builds on

33 of 43 references.

References · 33

  1. 2015Deep Residual Learning for Image Recognitioncited by 52,096
  2. 2021Learning Transferable Visual Models From Natural Language Supervisioncited by 33,744
  3. 2020An Image is Worth 16x16 Words: Transformers for Image Recognition at Scalecited by 30,186
  4. 2023Segment Anythingcited by 8,234
  5. 2023DINOv2: Learning Robust Visual Features without Supervisioncited by 7,974
  6. 2021Masked Autoencoders Are Scalable Vision Learnerscited by 7,403
  7. 2021Emerging Properties in Self-Supervised Vision Transformerscited by 6,968
  8. 2025VGGT: Visual Geometry Grounded Transformercited by 1,539
  9. 2024Depth Anything V2cited by 1,422
  10. 2025DINOv3cited by 1,414
  11. 2025SigLIP 2: Multilingual Vision-Language Encoders with Improved Semantic Understanding, Localization, and Dense Featurescited by 1,188
  12. 2016Semantic Understanding of Scenes through the ADE20K Datasetcited by 1,069
21 more
  1. 2020Learning Continuous Image Representation with Local Implicit Image Functioncited by 552
  2. 2019Stand-Alone Self-Attention in Vision Modelscited by 542
  3. 2018Recovering Realistic Texture in Image Super-resolution by Deep Spatial Feature Transformcited by 371
  4. 2024Probing the 3D Awareness of Visual Foundation Modelscited by 177
  5. 2023AM-RADIO: Agglomerative Vision Foundation Model -- Reduce All Domains Into Onecited by 169
  6. 2024StableNormal: Reducing Diffusion Variance for Stable and Sharp Normalcited by 154
  7. 2019CARAFE: Content-Aware ReAssembly of FEaturescited by 119
  8. 2024FeatUp: A Model-Agnostic Framework for Features at Any Resolutioncited by 116
  9. 2023CLIP-DINOiser: Teaching CLIP a few DINO tricks for open-vocabulary semantic segmentationcited by 96
  10. 2025MEt3R: Measuring Multi-View Consistency in Generated Imagescited by 95
  11. 2024OpenNeRF: Open Set 3D Neural Scene Segmentation with Pixel-Wise Features and Rendered Novel Viewscited by 69
  12. 2021High Fidelity Visualization of What Your Self-Supervised Representation Knows Aboutcited by 66
  13. 2023Learning to Upsample by Learning to Samplecited by 57
  14. 2024Denoising Vision Transformerscited by 40
  15. 2022SAPA: Similarity-Aware Point Affiliation for Feature Upsamplingcited by 32
  16. 2024LiFT: A Surprisingly Simple Lightweight Feature Transform for Dense ViT Descriptorscited by 27
  17. 2025LoftUp: Learning a Coordinate-Based Feature Upsampler for Vision Foundation Modelscited by 23
  18. 2025JAFAR: Jack up Any Feature at Any Resolutioncited by 21
  19. 2025Generalized Neighborhood Attention: Multi-dimensional Sparse Attention at the Speed of Lightcited by 18
  20. 2024A Refreshed Similarity-based Upsampler for Direct High-Ratio Feature Upsamplingcited by 7
  21. 2025FeatSharp: Your Vision Model Features, Sharpercited by 6

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2025: 5; 2026: 23.

28 papers.

Sort:

By area:

  1. 2025C3G: Learning Compact 3D Representations with 2K Gaussianscited by 21
  2. 2026INSID3: Training-Free In-Context Segmentation with DINOv3cited by 12
  3. 2025Upsample Anything: A Simple and Hard to Beat Baseline for Feature Upsamplingcited by 6
  4. 2025NAF: Zero-Shot Feature Upsampling via Neighborhood Attention Filteringcited by 6
  5. 2026Driving with DINO: Vision Foundation Features as a Unified Bridge for Sim-to-Real Generation in Autonomous Drivingcited by 5
  6. 2026HD-VGGT: High-Resolution Visual Geometry Transformercited by 3
  7. 2026FIRE3D: Feed-forward Interactive 3D Scene Reconstruction Within A Minutecited by 3
  8. 2026UPLiFT: Efficient Pixel-Dense Feature Upsampling with Local Attenderscited by 2
  9. 2025FaCT: Faithful Concept Traces for Explaining Neural Network Decisionscited by 1
  10. 2026Interpretable Motion-Attentive Maps: Spatio-Temporally Localizing Concepts in Video Diffusion Transformerscited by 1
  11. 2026Falcon Perceptioncited by 1
  12. 2026LESV: Language Embedded Sparse Voxel Fusion for Open-Vocabulary 3D Scene Understandingcited by 1
16 more
  1. 2026DINO Soars: DINOv3 for Open-Vocabulary Semantic Segmentation of Remote Sensing Imagerycited by 1
  2. 2025CountZES: Counting via Zero-Shot Exemplar Selectioncited by 0
  3. 2026Geometry without Position? When Positional Embeddings Help and Hurt Spatial Reasoningcited by 0
  4. 2026Spectral Probing of Feature Upsamplers in 2D-to-3D Scene Reconstructioncited by 0
  5. 2026DiveUp: Learning Feature Upsampling from Diverse Vision Foundation Modelscited by 0
  6. 2026dinov3.seg: Open-Vocabulary Semantic Segmentation with DINOv3cited by 0
  7. 2026SPAR: Single-Pass Any-Resolution ViT for Open-vocabulary Segmentationcited by 0
  8. 2026MapSR: Prompt-Driven Land Cover Map Super-Resolution via Vision Foundation Modelscited by 0
  9. 2026Frozen Vision Transformers for Dense Prediction on Small Datasets: A Case Study in Arrow Localizationcited by 0
  10. 2026Weighted Reverse Convolution for Feature Upsamplingcited by 0
  11. 2026ViT-Up: Faithful Feature Upsampling for Vision Transformerscited by 0
  12. 2026UNIEGO: Proxies as Mediators for Unified Egocentric Video Representation Learningcited by 0
  13. 2026RaysUp: Ultra-light Universal Feature Upsampling via Geometry-Aware Ray Representationcited by 0
  14. 2026From Reconstruction to Decision: A Post-Encoder Plug-in Adapter for Curvilinear Segmentationcited by 0
  15. 2026ORMA: Optimization-based Monocular 4D Reconstruction of Articulated Animalscited by 0
  16. 2026One Tile, Multiple Instances: Rethinking MIL for Sparse Diagnostic Evidencecited by 0