Relevant jobs
AI / Algorithm Engineer
QA Engineer
Data Analyst / Data Engineer
Video / Audio Producer
motion_reconstructionakanazawa/motion_reconstruction
Extract human keypoints from monocular video and reconstruct motion trajectories, providing usable pose data for motion …
No ratings yet
Rate this skill
Stars329
Forks72
Watchers329
Open Issues12
InstallOne-line install
UnconfirmedWhether this repository follows the Agent Skills spec (that is, whether it contains a SKILL.md) has not been verified yet. Until that is settled, clone it and use what you need:
git clone https://github.com/akanazawa/motion_reconstruction.git
OverviewOverview
Key Features
Extract human keypoints from monocular video and reconstruct motion trajectories, providing usable pose data for motion analysis and 3D pose research. Key capabilities include: 1. OpenPose Keypoint Extraction: run OpenPose frame by frame on input video via python -m run_openpose to output 2D joint keypoint sequences, removing the need to manually build a detection pipeline; 2. Video Pose Refinement: apply temporal smoothing and outlier correction to extracted keypoints via python -m refine_video, reducing the impact of jitter and missed detections on downstream reconstruction; 3. Reconstruction Pipeline Orchestration: chain keypoint extraction and refinement into a reproducible command-line pipeline that produces pose data under a unified input/output spec, ready to feed downstream 3D motion reconstruction models.
Use Cases
Monocular video human pose estimation, batch OpenPose keypoint extraction, data preprocessing before video motion reconstruction, pose sequence cleaning and smoothing, motion capture data generation.
Target Roles
AI / Algorithm Engineer, Data Analyst / Data Engineer, Film/TV Director / Audio-Video Producer, Test / QA Engineer.
RatingUser ratings
Nobody has rated this skill yet. Yours would be the first on the site — your vote becomes the initial average.
RepositoryRepository
| Skill ID | 2252 |
|---|---|
| Data source | GitHub |
| Repository | akanazawa/motion_reconstruction |
| Category | Computer Vision & Graphics / Human Pose & Motion Reconstruction |
| Primary language | Python |
| License | BSD 3-Clause "New" or "Revised" License |
| Default branch | master |
| GitHub repository ID | 194360295 |
| Node ID | MDEwOlJlcG9zaXRvcnkxOTQzNjAyOTU= |
| Rank | #1757 (position within the source ranking) |
| Site-wide popularity rank | #1974 (computed live from Stars, across all 5833 entries) |
GrowthStar / Fork growth
One snapshot per day, tracking how Stars and Forks changed
- Now329
- Previous point · 09-08 329
- vs previous point ±0
- 2 data points 35 days
- Now72
- Previous point · 09-08 72
- vs previous point ±0
- 2 data points 35 days
TimelineTimeline
Repository created2019-06-29 03:32:132019-06-29
Data last updated2026-10-26 12:22:1022 days ago
Last commit2022-09-16 18:32:082022-09-16
CommitLatest commit
Version [latest] committed by akanazawa. Update Category: Codebase Iteration & Routine Maintenance. Commit details: HMR to deepmimic skeleton. This update updates core functional modules and advances project iteration.