Sportiq

Sportiq performs player and ball tracking for tennis. It detects court boundaries, tracks players, and maps ball trajectories using Computer Vision (OpenCV), MediaPipe, and a TrackNet-based model.

Execution Guide:

The project has a FastAPI backend and a Vite React frontend. Use the root runner to start both in dev mode.

1. Clone the repository and navigate into it:

git clone https://github.com/kr1shnasomani/Sportiq.git
cd Sportiq

2. Start both servers:

./run_all.sh

What this does:

• Creates/uses a local Python virtual environment under backend/.venv and installs Python deps from backend/requirements.txt.
• Starts the FastAPI backend (auto-picks a free port like 8000/8001 and prints it, health at /health).
• Starts the Vite dev server for the frontend (auto-picks port like 8080/8081 and prints it, proxies /api to the backend).

Open the printed frontend URL in your browser, upload a tennis video, and download the processed result.

Note:

• The TrackNet weight file lives at backend/model/TrackNet.pth (already included here).
• Temporary files are handled automatically; you do not need to modify paths in code.

Run Services Individually (Optional):

Backend only:

cd backend
chmod +x run_backend.sh
PORT=8000 ./run_backend.sh

Frontend only:

cd frontend
chmod +x run_frontend.sh
VITE_BACKEND_URL="http://localhost:8000" VITE_DEV_PORT=8080 ./run_frontend.sh

API Usage (Direct):

The backend exposes a single main endpoint to process a video upload.

• Health check: GET /health
• Process video: POST /api/process (multipart/form-data, file field: video)

Example curl:

curl -X POST \
  -F "video=@/absolute/path/to/input.mp4" \
  http://localhost:8000/api/process \
  -o output.mp4

Output:

split.mp4

Workflow: