Workplace and physical security · Case study
Face-recognition entry and attendance system for offices
A face-recognition entry and attendance system for offices. A Raspberry Pi terminal recognizes staff with anti-spoofing checks, logs attendance from each entry, alerts on unknown faces and unauthorized attempts, and keeps running offline. Managers review access and attendance in a reporting dashboard.
The challenge
Cards and PINs get shared, lost or forgotten, and attendance records based on them are easy to game. The client wanted entry tied to the person, with a clear alert when someone unknown tried to get in.
Our approach
We built a face-recognition terminal on a Raspberry Pi using FaceNet embeddings, with anti-spoofing so a photo cannot open the door. The same check records attendance, unknown faces trigger alerts, and the device keeps working offline.
What we built
- Face recognition with anti-spoofing at the entry point
- Automatic attendance logging from each recognized entry
- Alerts for unknown faces and unauthorized entry attempts
- Offline operation on Raspberry Pi hardware
- Reporting dashboard for access and attendance
Highlights
- One check handles both access control and attendance
- Keeps working through network outages
Technology
- Python
- OpenCV
- FaceNet
- Raspberry Pi
- Flask
Access cards identify a card, not a person. We moved the check to the face: a camera on a Raspberry Pi matches each person against enrolled staff using FaceNet embeddings, and an anti-spoofing step stops someone holding up a photo.
Every recognized entry doubles as an attendance record, and unknown faces raise an alert straight away. Because recognition runs on the device, the door and the attendance log keep working when the network does not.
Client details are kept confidential. On a call we can walk you through this project in more depth, under NDA.