AI-Based Human Pose Estimation and Kinematic Analysis
Overview
The proposed service provides an automated pipeline for detection of human body posture and motion tracking from video recordings and for kinematic analysis of the performed movements. The pipeline receives video recordings of targeted activities and generates human-readable reports of kinematic analysis adapted to the end-user requirements. It supports clients that want to leverage the latest AI and HPC technology to improve their services or products that rely on measurements of human kinematics, particularly in the fields of sport science, ergonomics, and kinesiology. The service is based on existing state-of-the-art deep learning models for video-based human pose estimation and movement tracking, such as MediaPipe, Meta Sapiens, OpenPose, MoveNet, and MMPose. Video anonymization is implemented using established face detection and anonymization models. Existing models are augmented with additional labelled and synthetic training examples to increase robustness in problematic edge cases. The data processing pipeline supports GPU-accelerated execution and can be deployed on cloud or HPC infrastructure.
Outcome
The service facilitates high-precision kinematic measurement and analysis directly from standard video footage, obviating the need for costly or cumbersome motion-tracking hardware. It reduces equipment costs, simplifies workflow, and makes movement analysis accessible in diverse environments. Furthermore, the service generates curated, anonymized datasets, enabling the fine-tuning and tailoring of motion-tracking models to meet specific end-user requirements.
Service Card
| Category | AI products |
| Domain | Human Motion Analysis, Kinematics, Sports Science |
| Vertical | Health and Biotechnology |
| Target users | SMEs, Public Sector, Sports Scientists, Kinesiologists, Ergonomists |
| Service type | AI-Based Processing Pipeline |
| Outputs | Human-readable kinematic analysis reports; curated, anonymized datasets for model fine-tuning. |
| Requirements | |
| Limitations | |
| Availability | Q2/2027 |
| Licensing | |
| Visual content | assets/service-visuals/human-pose-estimation.png |
| Visual caption |