Edge AI

Awareness at the edge.

Smart security camera

On-device person detection makes camera alerts more useful while reducing video uploads.

YF / 05 Edge AI Smart security camera — conceptual system illustration
System concept Illustration / 05
Industry
Commercial security
Our role
Detection & edge deployment
Core tools
Jetson Nano · TensorRT · OpenCV
01 / The challenge

A practical problem.

Motion-only alerts created too many false alarms. Cloud video analysis added bandwidth costs and delayed notifications, so the client needed detection that could run on affordable local hardware.

Engagement
12 months · Q1 2023–Q1 2024
Project setting
Commercial security integrator · UK
02 / Our approach

Built around the constraints.

  1. Recognise people.

    A lightweight detector is trained for the client’s indoor and outdoor camera views.

  2. Optimise for the device.

    Quantisation and TensorRT reduce inference time on Jetson Nano.

  3. Send useful alerts.

    Local processing sends event metadata and alert thumbnails instead of continuous video uploads.

03 / The system

How it comes together.

Components → processing → output

Video stays close to the camera: local detection and tracking turn frames into alerts for the existing security system.

  1. Capture

    RTSP streams · USB cameras

    The capture module accepts camera feeds and negotiates resolution and frame rate for processing.

    OutputFrames for local inference

  2. Detect

    YOLOv7-tiny · TensorRT · Jetson Nano

    An optimised person detector runs on-device. Quantisation and layer fusion reduce the inference workload.

    OutputPerson detections & confidence

  3. Track

    DeepSORT · Multi-object tracking

    Detections are linked across frames to maintain individual tracks as people move through the camera view.

    OutputTracked people across frames

  4. Notify

    VMS integration · Alarm panels

    Zone-specific thresholds determine alerts. The system sends event metadata and thumbnails to existing monitoring tools.

    OutputAlerts, event metadata & thumbnails

04 / The outcome

Results from this project.

Person detection
99.2%Measured in daytime conditions
Typical inference
30 msDown from 45 ms on Jetson Nano
Less bandwidth
75%Compared with continuous cloud upload

Before

Motion triggers produced frequent false alarms; cloud analysis required continuous video uploads.

After

Local person detection sends focused alerts, event metadata, and thumbnails to the existing security system.

In context. Detection accuracy falls to 96.8% in nighttime infrared mode. Crowded scenes can increase inference time to approximately 45 ms.

Your next project

Working on something similar?

Tell us about your system and where you need support.

Discuss your project