MVP v2
This release contains the packaged, delivered product increment for the Assignment 4 Sprint. It introduces anti-spoofing mechanisms and hardware optimization for the FaceGuard system.
Required Links & Evidence
Changelog Summary
Added
- Liveness Detection to prevent spoofing via photos or videos
- Deployment to Raspberry Pi 5 and connection to the web-camera
- Camera selector on the Recognition page and Add Employee page
Changed
- Replaced buffalo_l model with a lightweight alternative for faster recognition and improved UI responsiveness
- Temporary registration now uses start and expiration date pickers instead of a number-of-days input
- Recognition page and Add Employee page: switched from single photo capture to video frame extraction
- Recognition session now starts with a Start button on the Recognition page
Artifacts & Deployment
The runnable artifact is provided via Docker Compose, supporting both arm64 (Raspberry Pi) and x86 (Laptops/PCs) architectures. Please refer to the Run Instructions linked above to build and start the containers.