The OpenWeedLocator (OWL) project provides an open-source, low-cost, image-based weed detection system for agricultural applications. It integrates a Raspberry Pi with a camera and relay control board to enable site-specific weed control, such as spot spraying. The project targets farmers, researchers, and hobbyists looking for an affordable and customizable solution for precision agriculture.

How It Works

OWL utilizes simple green-detection algorithms (e.g., Excess Green index, HSV thresholding) to identify weeds from camera imagery. The processed data triggers a relay control board, which can activate external devices like solenoid spray nozzles. The system is designed for flexibility, allowing integration with various Raspberry Pi models and cameras, and offers both 3D-printable and more robust aluminum enclosures.

Quick Start & Requirements

• Installation: A two-line bash script (bash ~/owl/owl_setup.sh) automates software installation on a Raspberry Pi with Raspbian OS. A detailed, step-by-step installation is also provided.
• Hardware: Requires a Raspberry Pi (3B+, 4, or 5 recommended), a compatible camera (Raspberry Pi Global Shutter Camera recommended), SD card, and power supply. Optional components include a Google Coral USB Accelerator for enhanced processing.
• Setup Time: Quick install is estimated at 10 minutes; detailed install at 60 minutes.
• Documentation: Comprehensive guides for hardware assembly, software installation, and configuration are available.

Highlighted Details

• Supports Raspberry Pi 5 with picamera2 and higher resolutions (640x480).
• Offers two enclosure designs: a simpler 3D-printable version and a more durable extruded aluminum "Compact OWL."
• Includes options for hardware controllers (Ute Controller, Advanced Controller) for managing multiple OWL units.
• Provides detailed troubleshooting guides for common hardware and software issues.

Maintenance & Community

The project is actively maintained, with recent updates including Raspberry Pi 5 support and improved enclosure designs. Community engagement is encouraged via the GitHub Discussions tab for ideas and feedback.

Licensing & Compatibility

The software is released under the MIT License, allowing for commercial use and modification. Hardware designs are also open-source.

Limitations & Caveats

The current green-detection algorithms are primarily effective in fallow conditions; "green-on-green" (in-crop) detection is in development and may require custom model training or specific hardware like a Google Coral accelerator. Performance can be affected by lighting conditions, with outdoor use being optimal.