mlsanityData to 14 Oct 2026

What Do Scan-Derived Class Prototypes Add? Disentangling Supervision, Prototype Content and Query Protocol in Recognition over Frozen Foundation Features

arXiv 2609.04381 ↗First posted 3 Sep 2026cs.CV cs.AI cs.ROcited by 1

Not on the map. Why?

Abstract (arXiv)

A scan supplies labeled images and a geometric reference. We separate their contributions in a recognizer whose scan-derived prototype matrix acts as a supervised head's fixed output layer. On T-LESS, HOPE and 18 self-collected industrial parts, we test real, random and exactly permuted prototypes, matched geometry-free classifiers, stronger appearance rules and paired background protocols. Across DINOv2-giant and MetaCLIP-H with real-background queries, the largest fused-accuracy advantage of the real prototypes over either control is one percentage point; larger differences favor controls, by up to 2.8 points in arm means. On HOPE with DINOv2-giant the head alone is 2.8 points above exact permutations (95% interval: 0.8-4.7); this advantage does not reach fusion and is not observed on MetaCLIP-H. On DINOv2-giant, matched logistic regression comes within 0.5 points of fusion on T-LESS and exceeds it on HOPE and the self-collected parts. Against white cutouts, real HOPE query backgrounds lower image-prototype accuracy by 43 points on DINOv2-giant and 13 on MetaCLIP-H. The audit separates prototype content, label supervision and query protocol.

Key facts. Cited by 1 paper in the map (data through 14 Oct 2026). Builds on 30 of its 40 references that are in the map. First posted on arXiv 3 Sep 2026.

Builds on

30 of 40 references.

References · 30

  1. 2021Learning Transferable Visual Models From Natural Language Supervisioncited by 33,744
  2. 2023DINOv2: Learning Robust Visual Features without Supervisioncited by 7,974
  3. 2021Emerging Properties in Self-Supervised Vision Transformerscited by 6,968
  4. 2016PointNet: Deep Learning on Point Sets for 3D Classification and Segmentationcited by 5,248
  5. 2017PointNet++: Deep Hierarchical Feature Learning on Point Sets in a Metric Spacecited by 4,401
  6. 2017Prototypical Networks for Few-shot Learningcited by 3,015
  7. 2025SigLIP 2: Multilingual Vision-Language Encoders with Improved Semantic Understanding, Localization, and Dense Featurescited by 1,188
  8. 2023Self-Supervised Learning from Images with a Joint-Embedding Predictive Architecturecited by 1,039
  9. 2019A Closer Look at Few-shot Classificationcited by 684
  10. 2018Recognition in Terra Incognitacited by 658
  11. 2021Point-BERT: Pre-training 3D Point Cloud Transformers with Masked Point Modelingcited by 653
  12. 2020Rethinking Few-Shot Image Classification: a Good Embedding Is All You Need?cited by 437
18 more
  1. 2023FoundationPose: Unified 6D Pose Estimation and Tracking of Novel Objectscited by 429
  2. 2020Noise or Signal: The Role of Image Backgrounds in Object Recognitioncited by 343
  3. 2020A Metric Learning Reality Checkcited by 321
  4. 2017T-LESS: An RGB-D Dataset for 6D Pose Estimation of Texture-less Objectscited by 210
  5. 2017Cut, Paste and Learn: Surprisingly Easy Synthesis for Instance Detectioncited by 209
  6. 2022MegaPose: 6D Pose Estimation of Novel Objects via Render & Comparecited by 201
  7. 2020BOP Challenge 2020 on 6D Object Localizationcited by 193
  8. 2023OpenShape: Scaling Up 3D Shape Representation Towards Open-World Understandingcited by 184
  9. 2022OnePose: One-Shot Object Pose Estimation without CAD Modelscited by 157
  10. 2023SparseGS: Sparse View Synthesis using 3D Gaussian Splattingcited by 134
  11. 2023GigaPose: Fast and Robust Novel Object Pose Estimation via One Correspondencecited by 114
  12. 2023FoundPose: Unseen Object Pose Estimation with Foundation Featurescited by 109
  13. 2023CNOS: A Strong Baseline for CAD-based Novel Object Segmentationcited by 87
  14. 2019Hyperspherical Prototype Networkscited by 86
  15. 2018Fix your classifier: the marginal value of training the last weight layercited by 43
  16. 2024Shape-biased Texture Agnostic Representations for Improved Textureless and Metallic Object Detection and 6D Pose Estimationcited by 2
  17. 2024A Coding-Theoretic Analysis of Hyperspherical Prototypical Learning Geometrycited by 1
  18. 2026Prior Availability in Industrial Visual Sim-to-Real: A Review of CAD-Guided and CAD-Unavailable Regimescited by 1

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2026: 1.

1 paper.

  1. 2026CALIPER: Metric-Grounded Model-Free Recognition of Visually Similar Industrial Partscited by 0