Gaze Detection

This project detects and visualizes the direction of a person's gaze in real-time using a webcam.

---

Features

• Real-time gaze direction detection: Classifies gaze as Left, Right, Up, Down, Center, or combinations (e.g., Up-Left).
• Visual feedback: Draws arrows from the iris centers indicating gaze direction, and overlays text on the video feed.
• Smoothing: Uses a short history of predictions to stabilize the output and reduce flicker.
• Robust iris tracking: Uses MediaPipe’s refined landmarks for accurate iris localization.

---

How It Works

1. Face & Iris Detection:
   MediaPipe FaceMesh detects facial landmarks, including iris positions.

2. Gaze Estimation:
   The code calculates the relative position of each iris within its eye using landmark ratios.

   • Horizontal and vertical ratios are computed for both eyes.
   • These ratios are classified into gaze directions (Left, Right, Up, Down, Center).

3. Smoothing:
   A short history of gaze predictions is kept. The most frequent recent prediction is used for display.

4. Visualization:

   • Arrows are drawn from each iris center in the direction of the detected gaze.
   • The current gaze direction is displayed as text.

---

Requirements

• Python 3.7+
• OpenCV (opencv-python)
• MediaPipe (mediapipe)

Install dependencies:

pip install opencv-python mediapipe

---

Usage

Run the application:

python app.py

• A window will open showing the webcam feed.
• Gaze direction is indicated by arrows and text.
• Press ESC to exit.

---

File Structure

Gaze-Detection/
├── app.py       # Main application code
├── README.md    # Project documentation

---

Customization

• Smoothing window:
  Change gaze_history = deque(maxlen=10) for more/less smoothing.
• Arrow length:
  Adjust length in draw_arrow() for longer/shorter arrows.
• Gaze thresholds:
  Modify values in get_gaze_direction() for sensitivity tuning.