‘AgriScout’ robot shows strong potential for PVY detection

An international team of researchers has developed and tested an autonomous robot named AgriScout that can detect Potato Virus Y (PVY) infections in potato crops with high precision. Their findings, published in the peer-reviewed journal Computers and Electronics in Agriculture, suggest that this technology could significantly improve disease scouting efficiency and accuracy in commercial potato production.

Field Deployment and Image Collection

AgriScout was deployed in commercial potato fields in New Brunswick and Prince Edward Island during the 2023 growing season. The robot is equipped with five Raspberry Pi cameras mounted on a foldable boom, each paired with LED lighting and shading curtains to ensure consistent image quality. Over 55,000 high-resolution RGB images of Russet potato plants were collected during peak symptom visibility in mid-July.

Researchers manually reviewed the images and confirmed 764 PVY-positive plants through laboratory testing. These images were used to train a deep learning model based on the YOLOv11 architecture.

What is YOLO?

YOLO stands for “You Only Look Once,” a real-time object detection algorithm widely used in computer vision. It processes images in a single pass, making it highly efficient for field applications where speed and accuracy are critical. In this study, the YOLOv11 model was trained using transfer learning and fine-tuned over 186 epochs to detect PVY symptoms in potato foliage.

Detection Accuracy and Mapping

The trained model achieved a mean Average Precision (mAP) of 85%, with a precision of 0.85 and recall of 0.76. In a separate test field, AgriScout flagged 123 suspected infections, 105 of which were confirmed PVY-positive—closely matching the expected 1.5% infection rate.

AgriScout uses GPS-RTK geolocation and Starlink satellite connectivity to geotag images and transmit them to a cloud server in real time. This enables the generation of infestation maps that pinpoint infected plants with centimeter-level accuracy, allowing growers to take targeted action.

Scalability and Global Collaboration

The research team includes contributors from Canada, the United States, and Australia—highlighting the global interest in advancing precision agriculture. While early-stage symptom detection remains a challenge, AgriScout’s modular design and robust performance under variable field conditions make it a promising tool for scalable disease management.

The researchers suggest that the system could be adapted for other crops and pathogens in future applications. For growers seeking to reduce labor costs and improve disease control, AgriScout offers a compelling glimpse into the future of smart potato farming.

Source: Computers and Electronics in Agriculture. Full study available here.
Related video: This robot is using AI to help P.E.I. farmers detect potato viruses. Credit CBC
Image: AgriScout working in a potato field. Credit CBC