Smart Home Face Recognition System

Developed a CNN-based face recognition system to control access in a smart home environment. The solution included:

Face Detection: Used mmod_human_face_detector from the dlib library.

Feature Extraction: Generated 128-dimensional facial embeddings using dlib’s embedding model.

Landmark Detection: Applied dlib’s 68-point shape predictor for precise facial landmarks.

Database Integration: Created and managed a database of authorized faces (myself and friends).

Access Control:

Recognized faces → green rectangle + automatic door unlock.

Unknown faces → red rectangle + “Trespasser” alert.

Owner options:

Allow entry without adding face to DB.

Allow entry & add face to DB.

Deny entry.

Hardware Integration: Connected to door system via Atmega32A microcontroller.

My Role: Implemented Python-based face recognition pipeline and designed the graphical user interface (GUI).