This project demonstrates real-time Face Landmark Detection using the MediaPipe Tasks API and OpenCV in Python.
It detects 468 facial landmark points from video frames using the face_landmarker.task model in VIDEO mode.
- ✅ Real-time face landmark detection
- ✅ 468 facial landmark points
- ✅ Video file & webcam support
- ✅ MediaPipe Tasks API (latest approach)
- ✅ Timestamp handling for VIDEO mode
- ✅ OpenCV visualization
- Python
- OpenCV
- MediaPipe Tasks API
- NumPy
- main1.py has low range effective accuracy for face mapping but feels faster due to higher frame rate
- main2.py has high range effective accuracy for face mapping but feels a bit slower due to low frame rate