Research map / 3D vision / Human pose estimation
Towards Stable Human Pose Estimation via Cross-View Fusion and Foot Stabilization
Li’an Zhuo, Jian Cao, Qi Wang, Bang Zhang, Liefeng Bo
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 |
|---|---|---|---|
| Replace ResNet50 with ViT-Base as the feature extraction encoder (ablation) | 3D pose estimator · 3DPW test | PA-MPJPE (mm): 48.5 -> 47.8 | Improved |
| Replace ViT-Base with ViT-Large as the feature extraction encoder (ablation) | 3D pose estimator · 3DPW test | PA-MPJPE (mm): 47.8 -> 44.5 | Improved |
| Add a front-view intermediate representation (ablation) | 3D pose estimator · 3DPW test | PA-MPJPE (mm): 48.7 -> 47.0 | Improved |
| Add a top-view representation to the front-view representation (ablation) | 3D pose estimator · 3DPW test | PA-MPJPE (mm): 47.0 -> 46.1 | Improved |
| Add a side-view representation without cross-view attention (ablation) | 3D pose estimator · 3DPW test | PA-MPJPE (mm): 46.1 -> 45.9 | Improved |
| Add cross-view attention to the three-view representation (ablation) | 3D pose estimator · 3DPW test | PA-MPJPE (mm): 45.9 -> 45.1 | Improved |
| Use a fixed root-to-leaf Kinematic Topology Decoder instead of independent joint prediction (ablation) | AIST++ pose estimator · AIST++ test | PA-MPJPE (mm), whole body: 37.5 -> 37.1 | Improved |
| Use contact-conditioned reversible kinematic topology instead of fixed KTD (ablation) | AIST++ pose estimator · AIST++ test | PA-MPJPE (mm), whole body: 37.1 -> 36.4 | Improved |
| Use contact-conditioned reversible kinematic topology instead of fixed KTD (ablation) | AIST++ pose estimator · AIST++ test | PA-MPJPE (mm), foot: 34.9 -> 33.6 | Improved |
Closest papers by meaning
- Consensus-Based Optimization for 3D Human Pose Estimation in Camera Coordinates (IJCV 2022 · similarity 0.86)
- Cross View Fusion for 3D Human Pose Estimation (ICCV 2019 · similarity 0.86)
- Deep Kinematics Analysis for Monocular 3D Human Pose Estimation (CVPR 2020 · similarity 0.85)
- SVTformer: Spatial-View-Temporal Transformer for Multi-View 3D Human Pose Estimation (AAAI 2025 · similarity 0.85)
- HumanMM: Global Human Motion Recovery from Multi-shot Videos (CVPR 2025 · similarity 0.85)
- Optimizing Human Pose Estimation Through Focused Human and Joint Regions (AAAI 2025 · similarity 0.84)
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.
