Raspberry Pi FarmGuard uses AI to detect troublesome wildlife and alert farmers

Raspberry Pi
(Image credit: Samith)

The Raspberry Pi has proven to be a competent board for creating various AI-powered projects. Today, we’ve got a great example of this technology in action known as FarmGuard, created by maker and developer Samith. With help from our favorite SBC, Samith can monitor farmland, detect unwanted wildlife, and notify users when something is spotted.

Samith said the goal was to create a system to help farmers look for animals that might accidentally damage their crops. The result is a monitor driven primarily by a Raspberry Pi. It’s paired with an AI camera for image processing and a Blues Notecard cellular IoT board for notifications.

The wildlife is identified using a module known as the Gravity HuskyLens Vision Sensor. This board has a camera and is ready to handle many AI tasks like object recognition. The module also sports a 2-inch IPS screen for video output. Samith trained the model with images of wild boar, elephants, and monkeys to test the setup.

Samith is using a Raspberry Pi 4 Model B for the project but there’s no reason you couldn’t get away with a Pi 5 if you’ve got one on hand. It’s working alongside the HuskyLens Vision Sensor and a Blues Notecard for SMS support. The Blues Notecard uses a special carrier board to connect to the Pi. All the hardware is held together using a case made from laser-cut acrylic sheets.

Raspberry Pi OS Bullseye Edition is the central operating system used by the Pi. It works with the notification system’s Twilio SMS Messaging API and Blues Notehub.io. You can read more about the code used in this Raspberry Pi project in greater detail at the official Hackster project page. There, you’ll also find a complete breakdown of the assembly process and instructions for anyone who wants to recreate it at home.

Ash Hill
Freelance News and Features Writer

Ash Hill is a Freelance News and Features Writer with a wealth of experience in the hobby electronics, 3D printing and PCs. She manages the Pi projects of the month and much of our daily Raspberry Pi reporting while also finding the best coupons and deals on all tech.

  • bit_user
    Here's the vision processor used by the HuskyLens Vision Sensor:
    https://www.canaan.io/product/kendryteai
    Reply
  • M0rtis
    I wonder if this could be used to make a home security paintball turret loaded with pepper spray or rubber balls
    Reply
  • bit_user
    M0rtis said:
    I wonder if this could be used to make a home security paintball turret loaded with pepper spray or rubber balls
    For moving targets, you're likely to have a latency problem. Another issue might be the minimum size object it can detect, as that's going to limit how much area you can cover with one camera.

    I've seen PTZ cameras with an auto-track feature, and they typically work by waiting until the subject reaches the edge of the image and then try to re-center (based on last observed position), rather than attempting to continuously follow it/them.
    Reply