Padex

Open-source Python toolkit that transforms padel match broadcast footage into structured tracking data, event streams, and tactical analytics.

padex = padel + x (analytics/extraction)

Installation

pip install padex

Quick Start

Python API

import padex

result = padex.process("match.mp4", calibration="match_calibration.json")
print(f"Detected {len(result.shots)} shots, {len(result.bounces)} bounces")

padex.export_video(result, "match.mp4", "output/annotated.mp4")

CLI

# Run full pipeline (auto-launches calibration if needed)
padex process match.mp4

# Calibrate court separately
padex calibrate match.mp4

Features

• Player Detection & Tracking -- YOLO + ByteTrack with team classification
• Ball Tracking -- TrackNet with Kalman filtering
• Court Calibration -- Interactive 12-point homography
• Bounce Detection -- Ground and wall bounce classification
• Shot Classification -- Pose-based three-signal decision tree (15 padel shot types)
• Video Annotation -- Annotated output with overlays, court lines, and shot labels

Documentation

Full documentation: padex.readthedocs.io

• Quick Start
• Pipeline Guide
• Shot Classification
• API Reference

Development

git clone https://github.com/rondo-labs/padex.git
cd padex
uv sync --group dev
uv run pytest tests/ -v

License

See LICENSE for details.