Research map / 3D vision / 3D generation
Towards A Richer 2D Understanding of Hands at Scale
Tianyi Cheng, Dandan Shan, Ayda Sultan Hassen, Richard Ely Locke Higgins, David Fouhey
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 proposed association mechanism instead of single-vector prediction of Shan et al. | hand-object association · Hands23, at least 6 hands | F1: 55.9 -> 79.1 | Improved |
| use proposed association mechanism instead of Shan et al. | hand-object association · Hands23, all | F1: 85.3 -> 89.9 | Improved |
| use proposed association mechanism | hand-object association · Hands23, all | recall: 93.9 -> 96.5 | Improved |
| use proposed association mechanism | hand detection · Hands23, at least 6 hands | hand detections count: 1.1K -> 2.3K | Improved |
| train on Hands23 and test zero-shot on other datasets instead of within-dataset training | Faster-RCNN · VLOG | AP: 80.9 -> 85.5 | Improved |
| train on Hands23 and test zero-shot instead of within-dataset training | Faster-RCNN · Ego | AP: 97.9 -> 96.4 | Hurt |
| train on Hands23 and test zero-shot instead of within-dataset training | Faster-RCNN · VGG | AP: 84.4 -> 64.4 | Hurt |
| train on Hands23 and test zero-shot instead of within-dataset training | Faster-RCNN · TV | AP: 71.0 -> 60.9 | Hurt |
| train on 100DOH instead of Hands23 and test on Hands23 | Faster-RCNN · Hands23 | AP: 85.6 -> 78.5 | Hurt |
| train on blurred faces and test on unblurred faces (ablation) | hand detector · Hands23 hand detection | AP: 84.3 -> 84.2 | No clear effect |
| train and test with blurred faces versus unblurred (ablation) | hand detector · Hands23 hand detection | AP: 84.3 -> 82.4 | No clear effect |
| self-training on Ego4D with SAM masks | proposed model · Ego4D val | Box AP: 86.9 -> 90.9 | Improved |
Closest papers by meaning
- Understanding Human Hands in Contact at Internet Scale (CVPR 2020 · similarity 0.92)
- OakInk: A Large-Scale Knowledge Repository for Understanding Hand-Object Interaction (CVPR 2022 · similarity 0.91)
- ContactPose: A Dataset of Grasps with Object Contact and Hand Pose (ECCV 2020 · similarity 0.90)
- How Do I Do That? Synthesizing 3D Hand Motion and Contacts for Everyday Interactions (CVPR 2025 · similarity 0.89)
- HandVQA: Diagnosing and Improving Fine-Grained Spatial Reasoning about Hands in Vision-Language Models (CVPR 2026 · similarity 0.87)
- GEARS: Local Geometry-aware Hand-object Interaction Synthesis (CVPR 2024 · similarity 0.87)
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.
