Research map / 3D vision / 3D generation
SHands: A Multi-View Dataset and Benchmark for Surgical Hand-Gesture and Error Recognition Toward Medical Training
Le Ma, Thiago Freitas dos Santos, Nadia Magnenat-Thalmann, Katarzyna Wac
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 |
|---|---|---|---|
| use transformer backbone (TimeSformer) instead of CNN (SlowFast) | video backbone · SHands single-view gesture recognition | Top-1 accuracy (%): 55.7 -> 58.4 | Improved |
| use masked pretraining (VideoMAE-L) instead of VideoMAE-B | VideoMAE · SHands single-view gesture recognition | Top-1 accuracy (%): 63.4 -> 65.9 | Improved |
| use InternVideo2 instead of VideoMAE-L | video backbone · SHands single-view gesture recognition | Top-1 accuracy (%): 65.9 -> 68.9 | Improved |
| use five views (DVANet) instead of single-view InternVideo2 | gesture recognition · SHands gesture recognition | Top-1 accuracy (%): 68.9 -> 73.9 | Improved |
| use DVANet instead of MVAction in multi-view | multi-view gesture recognition · SHands multi-view gesture recognition | Macro-F1: 0.704 -> 0.736 | Improved |
| use multi-view ViewCon instead of single-view VideoMAE-B | error detection · SHands error detection (trainee subset) | Top-1 accuracy: 60.4 -> 66.2 | Improved |
| use DVANet with all five views instead of single-view VideoMAE-B | error detection · SHands error detection (trainee subset) | Top-1 accuracy: 60.4 -> 68.5 | Improved |
| test on unseen views C4-C5 instead of seen views | ViewCon · SHands cross-view generalization | Top-1 accuracy (%): 68.4 -> 63.2 | Hurt |
| test on unseen views C4-C5 instead of seen views | DVANet · SHands cross-view generalization | Top-1 accuracy (%): 72.8 -> 68.5 | Hurt |
| use DVANet instead of ViewCon on unseen views | cross-view generalization · SHands cross-view generalization (unseen views) | Top-1 accuracy (%): 63.2 -> 68.5 | Improved |
Closest papers by meaning
- CholecTrack20: A Multi-Perspective Tracking Dataset for Surgical Tools (CVPR 2025 · similarity 0.87)
- HandVQA: Diagnosing and Improving Fine-Grained Spatial Reasoning about Hands in Vision-Language Models (CVPR 2026 · similarity 0.86)
- HandEdit: A Unified Benchmark for Egocentric Human-to-Robot Dexterous Hand Image Editing (NeurIPS 2026 · similarity 0.85)
- Towards Dynamic 3D Reconstruction of Hand-Instrument Interaction in Ophthalmic Surgery (NeurIPS 2025 · similarity 0.84)
- EgoExOR: An Ego-Exo-Centric Operating Room Dataset for Surgical Activity Understanding (NeurIPS 2025 · similarity 0.83)
- emg2pose: A Large and Diverse Benchmark for Surface Electromyographic Hand Pose Estimation (NeurIPS 2024 · similarity 0.83)
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.
