Research map / 3D vision / Human pose estimation
Occluded Human Body Capture with Frequency Domain Denoising Prior
Buzhen Huang, Chongyang Xu, Wentao Tang, Yuan Shu, Jingyi Ju, Binghui Zuo, Yangang Wang
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 pretrained frequency-domain denoising prior encoder to the 3D lifting branch (ablation) | Temporal Regression · OcMotion | MPJPE: 51.7 -> 49.1 | Improved |
| train 3D decoder with the same diffusion process via reprojected keypoints (ablation) | Prior model · OcMotion | MPJPE: 49.1 -> 48.5 | Improved |
| use DWT instead of DCT for denoising keypoints (ablation) | Denoised Keypoints · OcMotion | MPJPE: 51.6 -> 49.6 | Improved |
| use DCT instead of DWT for denoising keypoints (ablation) | Denoised Keypoints · OcMotion | Accel.: 18.8 -> 17.6 | Improved |
| add detected 2D keypoints from ViTPose (ablation) | Temporal Regression · OcMotion | MPJPE: 51.7 -> 52.3 | Hurt |
| add time-domain keypoint denoising with temporal transformer (ablation) | Temporal Regression · OcMotion | Accel.: 28.2 -> 23.3 | Improved |
| add DWT-based denoising of keypoints instead of time-domain denoising (ablation) | Denoised Keypoints · OcMotion | MPJPE: 51.4 -> 49.6 | Improved |
| add plain 3D diffusion lifting (ablation) | Temporal Regression + predicted keypoints · OcMotion | MPJPE: 52.3 -> 51.3 | No clear effect |
| apply DWT to 3D diffusion without pretrained 2D prior (ablation) | 3D Diffusion · OcMotion | MPJPE: 51.3 -> 51.0 | No clear effect |
| add ground-truth 2D keypoints (oracle) to temporal regression (ablation) | Temporal Regression · OcMotion | MPJPE: 51.7 -> 32.5 | Improved |
| replace frame-wise regression by temporal transformer regression (ablation) | Frame Regression · OcMotion | Accel.: 31.4 -> 28.2 | Improved |
| use frequency-domain method instead of CloseInt multi-person interaction method | Hi4D inter-person occlusion · Hi4D | MPJPE: 63.1 -> 61.5 | Improved |
Closest papers by meaning
- OccFusion: Rendering Occluded Humans with Generative Diffusion Priors (NeurIPS 2024 · similarity 0.87)
- DPMesh: Exploiting Diffusion Prior for Occluded Human Mesh Recovery (CVPR 2024 · similarity 0.87)
- RoHM: Robust Human Motion Reconstruction via Diffusion (CVPR 2024 · similarity 0.86)
- A Unified Diffusion Framework for Scene-aware Human Motion Estimation from Sparse Signals (CVPR 2024 · similarity 0.86)
- $\text{Di}^2\text{Pose}$: Discrete Diffusion Model for Occluded 3D Human Pose Estimation (NeurIPS 2024 · similarity 0.86)
- ODHSR: Online Dense 3D Reconstruction of Humans and Scenes from Monocular Videos (CVPR 2025 · similarity 0.86)
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.
