Skip to content
MAJ Technologies

Case study · Electronics · AI

AI home security: facial recognition at the door

A self-built access control system: camera, infrared night vision, on-device recognition and a phone alert when a stranger appears.

Client
Private installation
Published
15 July 2025
5 s to open the doorIR night vision~500 TND hardware cost

Challenge

Commercial access-control systems are expensive, cloud-locked and rarely built for Tunisian weather. The goal: a door that recognises family members day and night, opens fast, and costs a fraction of a commercial unit.

What we built

A camera module with infrared illumination for night vision, a Raspberry Pi running face recognition locally (no cloud round-trip) and a mobile app that shows every event. Known faces unlock the door in about 5 seconds; an unknown face triggers an alert with a snapshot.

The PCB and wiring were designed and assembled in-house, weather-sealed for outdoor use, including the rain test, which found (and fixed) a drainage issue.

Result

A working installation at roughly 500 TND in hardware, versus thousands for a commercial equivalent. All processing stays on the device, so there is no monthly fee and no video leaving the premises.

Stack

Raspberry PiPi Camera + IRPythonOpenCVMQTTMobile app
AI home security: facial recognition at the doorAI home security: facial recognition at the door

Tunis based. Working with Tunisia, the Gulf and Europe.

Tell us what you want to build. We reply within one business day.

---