Research map / 3D vision / Human pose estimation
DiffPose: SpatioTemporal Diffusion Model for Video-Based Human Pose Estimation
Runyang Feng, Yixing Gao, Tze Ho Elden Tse, Xueqing Ma, Hyung Jin Chang
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 LMSFI (lookup-based multi-scale feature interaction) to the direct concatenation baseline (ablation) | DiffPose · PoseTrack2017 val | mAP: 52.5 -> 82.4 | Improved |
| add spatiotemporal representation learner (STRL) as condition (ablation) | DiffPose with LMSFI · PoseTrack2017 val | mAP: 82.4 -> 86.4 | Improved |
| add multi-scale fusion over single low-resolution interaction (ablation) | LMSFI · PoseTrack2017 val | mAP: 84.4 -> 86.4 | Improved |
| use concatenation and aggregation instead of the default multi-scale fusion (ablation) | LMSFI · PoseTrack2017 val | mAP: 86.4 -> 86.2 | No clear effect |
| increase number of initial noises for pose ensemble from 1 to 5 to 10 (ablation) | DiffPose · PoseTrack2017 val | mAP: 80.4 -> 84.0 -> 86.4 | Improved |
| increase ensemble size N from 1 to 10 on challenging joints (ablation) | DiffPose · PoseTrack2017 val (wrist, ankle) | mAP: wrist 74.6 -> 83.5; ankle 74.1 -> 80.2 | Improved |
| increase denoising steps from 1 to 2 to 4 at inference without retraining (ablation) | DiffPose · PoseTrack2017 val | mAP: 85.9 -> 86.3 -> 86.4 | Improved |
| increase temporal interval delta from 0 to 1 to 2 (ablation) | STRL · PoseTrack2017 val | mAP: 82.4 -> 84.5 -> 86.4 | Improved |
| replace deterministic pose estimator FAMI-Pose with DiffPose diffusion framework | video pose estimation · PoseTrack21 | mAP: 81.2 -> 82.9 | Improved |
Closest papers by meaning
- Disentangled Diffusion-Based 3D Human Pose Estimation with Hierarchical Spatial and Temporal Denoiser (AAAI 2024 · similarity 0.90)
- $\text{Di}^2\text{Pose}$: Discrete Diffusion Model for Occluded 3D Human Pose Estimation (NeurIPS 2024 · similarity 0.90)
- Object Pose Estimation via the Aggregation of Diffusion Features (CVPR 2024 · similarity 0.89)
- Video-Based Human Pose Regression via Decoupled Space-Time Aggregation (CVPR 2024 · similarity 0.89)
- DiffusionRegPose: Enhancing Multi-Person Pose Estimation using a Diffusion-Based End-to-End Regression Approach (CVPR 2024 · similarity 0.88)
- Optimizing Human Pose Estimation Through Focused Human and Joint Regions (AAAI 2025 · similarity 0.88)
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.
