Live Camera Tracking
Real-time face detection and landmark tracking using your webcam
Interactive Demo
A p5.js simulation of face detection with landmarks, bounding boxes, and mask overlays
Live Recording
Actual face tracking with real-time mask overlay on webcam input
Recording of the face tracking system detecting facial landmarks and applying a dynamic mask in real-time
About This Project
Computer vision meets creative expression
This project explores real-time face tracking and mask overlays using computer vision and machine learning. It detects 68 facial landmarks in live video and applies dynamic masks that follow the user's face movements, expressions, and head rotations in real-time. The system uses a multi-stage pipeline: first a convolutional neural network localizes the face within the frame, then a regression model precisely maps landmark positions for eyes, eyebrows, nose, mouth, and jawline. The mask overlay system uses these landmarks as anchor points, interpolating between them to create smooth, natural-looking augmented reality effects.
Face Detection
Real-time face localization using a CNN-based detector with high accuracy bounding box regression.
Landmark Detection
68-point facial landmark detection mapping eyes, nose, mouth, jawline, and eyebrow positions.
Mask Overlays
Dynamic mask rendering anchored to facial landmarks with real-time expression tracking.
Motion Tracking
Smooth head pose estimation and movement tracking for natural mask alignment.