SensoryPi — Smart Security System (Raspberry Pi)
Built a Raspberry Pi-based smart security system that combines facial recognition with motion detection, triggering alerts on unrecognized faces and logging events with timestamped images.
Highlights
- Implemented real-time facial recognition using OpenCV face encodings to distinguish known from unknown visitors
- Integrated a PIR motion sensor via Arduino to trigger the camera pipeline on detected movement
- Built an event logging system that captures, timestamps, and stores alert frames for review
- Designed a hardware-software bridge between Arduino sensor input and Python/OpenCV processing on the Pi
Impact
- End-to-end embedded security system running on commodity hardware — camera, sensor, compute, and alerting in one device
- Demonstrated how real-time computer vision can be deployed at the edge with constrained resources
- Practical foundation in hardware-software integration: sensor input → processing → output action, the pattern that underlies modern IoT pipelines
Context
Physical security systems rely on two primitives: detecting that something moved and deciding if that something belongs there. This project implements both on edge hardware.
What I Built
A hardware-integrated smart security system running on a Raspberry Pi that:
- Detects motion via a PIR sensor connected through Arduino — no wasted compute polling an idle camera feed
- Activates the camera pipeline on a positive motion signal
- Runs facial recognition using OpenCV — comparing detected faces against a set of known enrollments
- Triggers an alert and logs a timestamped image frame when an unrecognized face is detected
The Arduino handles low-level sensor I/O; the Pi handles vision processing. The two communicate over serial, keeping the architectures clean.
Outcomes
- A functional security system running on a Raspberry Pi that correctly identifies known faces and alerts on unknowns
- Hardware-software integration spanning two embedded platforms (Arduino + Pi) and a computer vision stack
- A real-time system constrained by edge resources — frame rate, memory, and latency all matter
Why This Matters
Edge-deployed computer vision is an increasingly common pattern in industrial IoT, physical security, and smart infrastructure. Building one from hardware up — sensor, controller, compute, storage, alert — is fundamentally different from calling an API. This project built that foundation.