Research map / 3D vision / Human pose estimation
DiffPose: Toward More Reliable 3D 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 DiffPose reverse-diffusion pipeline instead of single-step pose estimation (ablation) | DiffPose · Human3.6M video-based, detected 2D poses | MPJPE: 44.3 -> 36.9 | Helped |
| Use reverse diffusion rather than repeated direct pose prediction (ablation) | DiffPose · Human3.6M video-based, detected 2D poses | MPJPE: 41.1 -> 36.9 | Helped |
| Use GMM-based forward diffusion instead of standard diffusion (ablation) | DiffPose · Human3.6M video-based, detected 2D poses | MPJPE: 40.1 -> 36.9 | Helped |
| Increase GMM kernels from 1 to 5 (ablation) | DiffPose · Human3.6M video-based, detected 2D poses | MPJPE: 38.0 -> 36.9 | Helped |
| Increase GMM kernels from 5 to 9 (ablation) | DiffPose · Human3.6M video-based, detected 2D poses | MPJPE: 36.9 -> 36.5 | Helped |
| Condition reverse diffusion on spatial-temporal context features (ablation) | DiffPose · Human3.6M video-based, detected 2D poses | MPJPE: 42.1 -> 39.3; 40.9 -> 36.9 | Helped |
| Increase reverse diffusion steps up to 50 (ablation) | DiffPose · Human3.6M video-based, detected 2D poses | MPJPE | Helped |
| Increase reverse diffusion steps beyond 50 (ablation) | DiffPose · Human3.6M video-based, detected 2D poses | MPJPE | No clear effect |
| Increase the number of sampled poses up to 5 (ablation) | DiffPose · Human3.6M video-based, detected 2D poses | MPJPE | Helped |
| Increase the number of sampled poses beyond 5 (ablation) | DiffPose · Human3.6M video-based, detected 2D poses | MPJPE | No clear effect |
| Add DDIM inference acceleration (ablation) | DiffPose · Human3.6M video-based, detected 2D poses | FPS: 173 -> 671 | Helped |
| Add DDIM inference acceleration (ablation) | DiffPose · Human3.6M video-based, detected 2D poses | MPJPE: 36.7 -> 36.9 | Hurt |
Closest papers by meaning
- $\text{Di}^2\text{Pose}$: Discrete Diffusion Model for Occluded 3D Human Pose Estimation (NeurIPS 2024 · similarity 0.92)
- Dual-Diffusion for Binocular 3D Human Pose Estimation (NeurIPS 2024 · similarity 0.90)
- Object Pose Estimation via the Aggregation of Diffusion Features (CVPR 2024 · similarity 0.90)
- MonoDiff: Monocular 3D Object Detection and Pose Estimation with Diffusion Models (CVPR 2024 · similarity 0.90)
- Disentangled Diffusion-Based 3D Human Pose Estimation with Hierarchical Spatial and Temporal Denoiser (AAAI 2024 · similarity 0.89)
- 6D-Diff: A Keypoint Diffusion Framework for 6D Object 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.