Sign Language Detector (Prototype)

A minimal, real-time sign detector using MediaPipe Hands and OpenCV. Classifies a small subset of gestures: numeric finger counts (0–5) and the "I Love You" (ILY) sign.

Features

• Real-time hand landmark tracking (MediaPipe Hands)
• Simple rule-based classification for: 0–5 and ILY
• On-screen overlays with FPS and detected labels

Requirements

• Python 3.9–3.12
• macOS with camera permission granted to your terminal/VS Code

Setup

python3 -m venv .venv
source .venv/bin/activate
pip install -r requirements.txt

Run

python hackathon.py --camera 0 --width 1280 --height 720

Press q to quit.

Notes

• This is an educational prototype, not a full ASL interpreter.
• Numeric labels 0–5 are approximate via finger-count heuristics and may not exactly match ASL number forms.
• The ILY heuristic looks for thumb, index, and pinky extended with middle and ring folded.
• Lighting, camera angle, and handedness affect accuracy.

Troubleshooting

• If the camera does not open on macOS, grant camera permissions to your terminal app or VS Code.
• Try --camera 1 if you have multiple cameras.