HUB
Sign inSign up

samymetref/football-match-analyzer

By samymetref

•Updated 8 months ago

Multi-Agent & AI-powered football match analysis — tracking, events, stats & reports via Telegram

Image
Machine learning & AI
0

449

samymetref/football-match-analyzer repository overview

⁠⚽ Football Match Analyzer

Multi-agent AI system that watches football match footage and generates professional-grade analytics — passes, tackles, physical stats, match reports — all from a Telegram bot.

6 AI agents. 1 Telegram bot. $0 cloud cost.

⁠Quick Start

# 1. Create .env with your keys (both free)
echo "GROQ_API_KEY=your_key" > .env
echo "GROQ_MODEL=llama-3.3-70b-versatile" >> .env
echo "LLM_PROVIDER=groq" >> .env
echo "TELEGRAM_BOT_TOKEN=your_token" >> .env

# 2. Create a data folder for match videos
mkdir data

# 3. Run
docker run --env-file .env -v $(pwd)/data:/app/data samymetref/football-match-analyzer:latest

Place an MP4 match video in your local data/ folder, then message the Telegram bot to analyze.

⁠What's Inside

  • YOLOv8s + ByteTrack — player detection & tracking across thousands of frames
  • KMeans on HSV — automatic team classification from jersey colors
  • Rule-based event detection — passes, tackles, interceptions
  • Physical stats engine — distance, top speed, sprints, heatmaps per player
  • Groq LLM reports — natural language match analysis (Llama 3.3 70B, free tier)
  • LangGraph + LangChain — multi-agent orchestration with tool calling
  • Token-efficient architecture — ~2K tokens/message, runs on free APIs

⁠Get Your Keys

Tag summary

Content type

Image

Digest

sha256:7ddaeffbd…

Size

581 MB

Last updated

8 months ago

docker pull samymetref/football-match-analyzer