Research map / 3D vision / Human pose estimation
Occlusion-Aware Networks for 3D Human Pose Estimation in Video
Yu Cheng, Bo Yang, Bo Wang, Wending Yan, Robby T. Tan
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 |
|---|---|---|---|
| add 2D TCN (ablation) | 3D TCN baseline · Human3.6M Protocol #1 | MPJPE (mm): 54.0 -> 51.7 | Improved |
| add occlusion awareness (ablation) | 3D TCN + 2D TCN · Human3.6M Protocol #1 | MPJPE (mm): 51.7 -> 46.3 | Improved |
| add occlusion awareness (ablation) | 3D TCN + 2D TCN · Human3.6M Protocol #2 | P-MPJPE (mm): 40.5 -> 35.4 | Improved |
| add symmetry constraint (ablation) | occlusion-aware TCNs · Human3.6M Protocol #1 | MPJPE (mm): 46.3 -> 45.8 | Improved |
| add adversarial learning (ablation) | occlusion-aware TCNs · Human3.6M Protocol #1 | MPJPE (mm): 45.8 -> 45.1 | Improved |
| add pose regularization (ablation) | occlusion-aware TCNs · Human3.6M Protocol #1 | MPJPE (mm): 45.1 -> 44.8 | Improved |
| add virtual view-angle data augmentation (ablation) | occlusion-aware TCNs · Human3.6M Protocol #1 | MPJPE (mm): 44.8 -> 42.9 | Improved |
| increase sequence length from 16 to 128 (ablation) | TCNs · Human3.6M Protocol #1 | MPJPE (mm): 55.4 -> 42.9 | Improved |
| increase sequence length from 128 to 256 (ablation) | TCNs · Human3.6M Protocol #1 | MPJPE (mm): 42.9 -> 44.1 | Hurt |
| raise reliability threshold from 0.30 to 0.40 (ablation) | TCNs · Human3.6M Protocol #1 | MPJPE (mm): 42.9 -> 44.2 | Hurt |
| lower reliability threshold from 0.30 to 0.20 (ablation) | TCNs · Human3.6M Protocol #1 | MPJPE (mm): 42.9 -> 45.7 | Hurt |
| use occlusion-aware framework instead of Pavllo et al. | 3D pose estimation · Human3.6M Protocol #1 | MPJPE (mm): 46.8 -> 42.9 | Improved |
Closest papers by meaning
- 3D Human Pose Estimation Using Spatio-Temporal Networks with Explicit Occlusion Training (AAAI 2020 · similarity 0.93)
- Human Pose Recognition via Occlusion-Preserving Abstract Images (ECCV 2024 · similarity 0.92)
- Explicit Occlusion Reasoning for Multi-Person 3D Human Pose Estimation (ECCV 2022 · similarity 0.91)
- 3D-Aware Neural Body Fitting for Occlusion Robust 3D Human Pose Estimation (ICCV 2023 · similarity 0.91)
- VOccl3D: A Video Benchmark Dataset for 3D Human Pose and Shape Estimation under real Occlusions (ICCV 2025 · similarity 0.90)
- $\text{Di}^2\text{Pose}$: Discrete Diffusion Model for Occluded 3D Human Pose Estimation (NeurIPS 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.
