Research map / 3D vision / Human pose estimation
Benchmarks and Challenges in Pose Estimation for Egocentric Hand Interactions with Objects
Zicong Fan, Takehiko Ohkawa, Linlin Yang, Nie Lin, Zhishan Zhou, Shihao Zhou, Jiajun Liang, Zhong Gao, Xuanyang Zhang, Xue Zhang, Fei Li, Liu Zheng and 12 more
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 perspective crop with virtual camera to reduce fisheye-rectified distortion | AssemblyHands hand pose methods · AssemblyHands test | MPJPE (mm): baseline 20.31 -> 24.85 from center to edge; method 1 stays 12.35 -> 13.72 | Improved |
| average four egocentric views instead of two lower views (ablation) | AssemblyHands baseline · AssemblyHands test | MPJPE (mm): 21.52 -> 20.69 | Improved |
| fuse multiple views by simple average instead of single view (ablation) | AssemblyHands baseline · AssemblyHands test | MPJPE (mm): 22.19 -> 21.52 | Improved |
| use adaptive view selection and high-capacity backbone (Hiera, RegNetY320) in place of ResNet50 baseline | AssemblyHands · AssemblyHands test | MPJPE (mm): 20.69 -> 12.21 and 12.46 | Improved |
| scale frozen ViT backbone from ViT-L to ViT-G (ablation) | transformer decoder method on ARCTIC · ARCTIC allocentric validation | CDev (mm): 30.5 -> 29.0 | Improved |
| replace MLP regressors with transformer decoder and frozen DINOv2 ViT-G backbone | ArcticNet-SF · ARCTIC allocentric test | CDev (mm): 41.56 -> 27.97 | Improved |
| add Swin-L Deformable DETR method with incomplete training | ArcticNet-SF · ARCTIC allocentric test | CDev (mm): 41.56 -> 64.15 | Hurt |
| evaluate from egocentric view rather than allocentric view | ARCTIC methods · ARCTIC test | hand pose error vs object error | Mixed |
Closest papers by meaning
- EgoPHI: Estimating 3D Hand-Object Contact and Force from Egocentric Vision (ECCV 2026 · similarity 0.91)
- AssemblyHands: Towards Egocentric Activity Understanding via 3D Hand Pose Estimation (CVPR 2023 · similarity 0.91)
- HOT3D: Hand and Object Tracking in 3D from Egocentric Multi-View Videos (CVPR 2025 · similarity 0.91)
- EgoPressure: A Dataset for Hand Pressure and Pose Estimation in Egocentric Vision (CVPR 2025 · similarity 0.90)
- MEgoHand: Multimodal Egocentric Hand-Object Interaction Motion Generation (NeurIPS 2025 · similarity 0.90)
- Single-to-Dual-View Adaptation for Egocentric 3D Hand Pose Estimation (CVPR 2024 · similarity 0.89)
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.
