Version 2.0 extends the embedded vision system by adding real-time face recognition and identity management.

New capabilities include:

• Face recognition using 128D embeddings (face_recognition / dlib)
• Automatic dataset loading at startup
• Runtime enrollment of unknown faces
• Stable single-face recognition tracking
• Modular vision pipeline architecture
• Architecture diagram and documentation improvements

The system maintains the lightweight event-driven design introduced in V1, ensuring efficient operation on Raspberry Pi hardware.

Key pipeline stages:

Camera → Motion Detection → Face Detection → Face Cropping → Embedding Generation → Face Recognition → Optional Enrollment → Visualization + TTS
Version 2.0 extends the embedded vision system by adding real-time face recognition and identity management.

This version demonstrates how classical computer vision and pretrained ML models can be integrated into a practical embedded system.

Initial release of the Embedded Vision System.

Features:

• Motion detection using frame differencing
• Face detection using OpenCV Haar cascades
• Event-driven processing to reduce CPU usage
• Text-to-speech alerts using pyttsx3