SIT Face Recognition System and Photobooth
A dual-mode face-recognition and event photobooth system for SIT
A real-time PyQt6 desktop system that combines InsightFace recognition, offline voice greetings, framed event photography, server upload, and QR-code sharing.
The 15-second brief
At a glance
- Outcome
- Unified identity recognition and photobooth workflows in one desktop runtime while releasing camera and inference resources when recognition is not required.
- Delivery
- Shipped, 2025
- Focus
- Systems, ML
- Scope
- 7 core technologies
Ownership and evidence
What I built
The product decisions and engineering work that shaped the final result.
Separated the PyQt6 interface, orchestration, face-matching rules, infrastructure, and data concerns through a five-layer architecture with one-directional dependencies.
Kept the 720p camera preview responsive by separating camera capture, recognition, and text-to-speech work across threads.
Precomputed face embeddings for fast matching and exposed recognition confidence controls for event operators.
Built a three-second photobooth capture flow with four selectable frame overlays, FastAPI upload, and an immediate QR code for sharing the result.
Product evidence
See it in action
Interface views and demonstrations from the working product.
Visual assets are not published for this project. The verified delivery and engineering scope are documented above.