Real-time face detection and facial landmark extraction system built using MediaPipe Face Mesh and OpenCV. The project supports image, video, and live webcam processing, enabling high-precision 3D landmark tracking (468 points) for each face. Designed with a modular pipeline, it allows efficient landmark extraction, normalization, and conversion into structured datasets for downstream machine learning tasks such as emotion recognition. Combines computer vision and ML-ready feature engineering to bridge real-time perception with intelligent applications.