Back

PlyoTracker AI

AI Analysis of Plyometric Training — UXV Center

PlyoTracker AI was built during a 4-month internship at UXV Center GmbH (Switzerland, remote). It analyzes plyometric football training from live video: MediaPipe pose estimation and TensorFlow/Keras classify exercises in real time (25–30 FPS) and compute biomechanical metrics (jump height, flight time, ground contact time, stability). The FastAPI backend handles JWT auth with coach/athlete roles; the React/TypeScript dashboard shows live metrics, session history, per-player group statistics and generates PDF reports.

7

Technologies

4

Key Features

Technologies Used

  • Python
  • TensorFlow
  • MediaPipe
  • OpenCV
  • FastAPI
  • React
  • TypeScript
PlyoTracker AI

Key Features

  • Real-time pose estimation at 25–30 FPS on live video streams
  • Live biomechanics: jump height, flight time, GCT, stability score
  • Coach / athlete roles with JWT authentication
  • Group mode with per-player statistics and automated PDF reports