1. Input: Face/Eye Image or Video (RGB/NIR)

2. Detection & Localization (Face, Eye Region, Pupil)

3. Head Pose Estimation (Roll, Pitch, Yaw)

4. Feature Extraction

Geometric Features (Pupil, Glints)

Appearance Features (Raw Pixels)

5. Model-Based Estimation (Geometric/Physiological Model)

6. Appearance-Based Estimation (DNN/CNN)

7. Output: 2D or 3D Gaze Direction Vector