AI / Computer Vision / Sports Technology
Pose Estimation
A thesis project applying computer vision and machine learning to analyse athlete movement from standard video input. The system extracts joint positions frame-by-frame using pose estimation, producing interpretable movement data for coaches and analysts — surfacing patterns the human eye misses under pressure or speed.
What it does
- Real-time joint tracking using MediaPipe pose estimation
- Movement pattern analysis from standard video input
- Body, leg, ankle, and movement attribute extraction per frame
- Injury-risk signal identification and flagging for coaches
- Readable output designed for non-ML sports professionals
Thesis / Research