Research map / Detection and segmentation / CNN features
Harmonious Feature Learning for Interactive Hand-Object Pose Estimation
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 the HFL-Net backbone and feature interaction modules | Joint hand-object pose estimation · HO3D | Joint: 10.1 -> 8.9 | Helped |
| Use the HFL-Net backbone and feature interaction modules | Joint hand-object pose estimation · HO3D | ADD-0.1D: 67.7 -> 73.3 | Helped |
| Use the HFL-Net backbone and feature interaction modules | Joint hand-object pose estimation · Dex-YCB s0 | MPJPE: 15.27 -> 12.56 | Helped |
| Use the HFL-Net backbone and feature interaction modules | Joint hand-object pose estimation · Dex-YCB s0 | ADD-0.1d(s): 29.8 -> 30.2 | Helped |
| Use independent middle stages instead of one shared stream (ablation) | HFL-Net backbone · HO3D | Joint: 10.4 -> 9.8 | Helped |
| Use shared low- and high-level stages with independent middle stages instead of separate full streams (ablation) | HFL-Net backbone · HO3D | Joint: 9.7 -> 9.8 | Hurt |
| Add interaction modules to the HFL-Net backbone (ablation) | HFL-Net · HO3D | Joint: 9.8 -> 8.9 | Helped |
| Use hand-stream features rather than object-stream features as attention keys and values (ablation) | Hand-to-object enhancement · HO3D | average ADD-0.1D: 69.5 -> 71.3 | Helped |
| Add hand-to-object feature enhancement (ablation) | HFL-Net · HO3D | average ADD-0.1D: 67.5 -> 71.3 | Helped |
| Add object-to-hand enhancement using self-attention on concatenated hand and object features (ablation) | HFL-Net · HO3D | Joint: 9.6 -> 8.9 | Helped |
| Concatenate hand and object features instead of adding them elementwise before self-attention (ablation) | Object-to-hand enhancement · HO3D | Joint: 9.3 -> 8.9 | Helped |
| Use self-attention on hand features alone (ablation) | Object-to-hand enhancement · HO3D | Joint: 9.6 -> 9.5 | Helped |
Closest papers by meaning
- Collaborative Learning for Hand and Object Reconstruction With Attention-Guided Graph Convolution (CVPR 2022 · similarity 0.90)
- UniHOPE: A Unified Approach for Hand-Only and Hand-Object Pose Estimation (CVPR 2025 · similarity 0.89)
- DeepHand: Robust Hand Pose Estimation by Completing a Matrix Imputed With Deep Features (CVPR 2016 · similarity 0.89)
- HOPE-Net: A Graph-Based Model for Hand-Object Pose Estimation (CVPR 2020 · similarity 0.89)
- HOISDF: Constraining 3D Hand-Object Pose Estimation with Global Signed Distance Fields (CVPR 2024 · similarity 0.87)
- SiMHand: Mining Similar Hands for Large-Scale 3D Hand Pose Pre-training (ICLR 2025 · 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.