01
Body detection
HomeGuard first identifies the person in frame so the system can focus on human movement instead of background motion.
YOLOv8How it works
HomeGuard turns a fall into a private, immediate family alert.
HomeGuard is a standalone project in the Aegis family. This page introduces the concept: a home camera runs edge AI locally, detects fall patterns, and sends only the alert evidence a trusted family member needs.
02
Pose tracking
The system extracts key body joints and follows posture changes over time, including torso direction and limb position.
MediaPipe
03
Motion analysis
Acceleration, body angle, and sudden posture changes are compared to understand whether the movement resembles a fall.
OpenCV
04
Fall classified
!
Fall detected
When a fall is classified, HomeGuard prepares an encrypted alert and a short event clip for authorized family review.
Smart decision enginePrivacy-first by design
Analysis runs on-device using edge AI. Raw footage stays inside the home unless a real alert needs review.
Secure family notifications
Only encrypted event clips and alert details are shared with authorized family members.
Built for real homes
The workflow is designed around practical movement signals: person detection, pose change, acceleration, and body angle.