An Edge AI–based real-time person detection and email alert system built using Raspberry Pi 5 and Hailo AI for forest surveillance and anti-poaching.

Community Projects
, , , ,
1

**Poacher Detection & Email Alert System using Raspberry Pi 5 and Hailo AI
**
Hello Hailo Community,

I’m sharing a completed Edge AI project built using Raspberry Pi 5 and the Hailo AI accelerator, focused on real-time human detection and alerting in restricted or remote areas such as forests.

The motivation behind this project was to design a low-power, portable surveillance system that can operate entirely at the edge without relying on cloud inference, making it suitable for deployment in locations with limited connectivity.

Hardware Used:

  1. Raspberry Pi 5
  2. Hailo AI Accelerator (Hailo-8 / Hailo-8L)
  3. Raspberry Pi Camera Module
  4. GPIO-based alert output (LED / buzzer)

Software & Frameworks:

  1. Hailo TAPPAS SDK
  2. GStreamer-based detection pipeline
  3. Python
  4. OpenCV
  5. SMTP (email alerts using app password authentication)

How the system works:
The application runs a real-time detection pipeline using Hailo’s basic_pipelines/detection.py.

Detection Flow:

  • Live camera frames are processed on-device using the Hailo accelerator

  • When a person is detected with confidence greater than 96%, the system waits for 2 seconds to confirm persistence

  • A snapshot is captured with bounding box and timestamp

  • The snapshot is automatically sent via email as an alert

  • A 10-minute cooldown is applied to avoid repeated alerts

  • A GPIO pin is triggered to support physical alerts

  • All snapshots are stored locally for later review

Key Features:

  • Real-time person detection using Edge AI

  • No cloud dependency for inference

  • Automated email alert with image attachment

  • Cooldown mechanism to prevent alert spamming

  • GPIO-based alert trigger

  • Low power and portable deployment

  • Can be configured to run automatically on boot

**Use Cases:
**
Although designed for forest surveillance and anti-poaching, the system can also be adapted for:

  • Border and defense surveillance

  • Wildlife monitoring

  • Perimeter security

  • Restricted-area intrusion detection

  • Remote installations with limited connectivity

GitHub Repository:

Future Enhancements:

  • Pose-based animation and educational applications

  • Bluetooth-based Wi-Fi configuration for remote deployment

  • Solar-powered operation for long-term field use

  • Multi-class detection and alert prioritization

Feedback and suggestions from the community are welcome.
Thanks for checking out the project!

Media:

image
image1081×1920 358 KB

Hardware - RaspberryPi 5 with Hailo 8(L) AI accelerator

image
image800×281 21 KB

Terminal Screenshot

image
image782×676 41.3 KB

Email Screenshot

3 Likes
2

Hi @Avaneesh

Thank you for sharing this comprehensive, real-world end-to-end edge AI application! It’s truly impressive to see such a fully working use case that demonstrates the capabilities of advanced edge AI. We’re thrilled that you’ve achieved these results, and we’re here to support you with any next steps or future developments.

1 Like