Soybean farmers may soon be able to identify crop diseases in the field using nothing more than a cellphone and an artificial intelligence-powered diagnostic tool developed by the University of Maryland's Digital and Precision Agriculture Lab.
The lab has released the first phase of SoyAI for field testing. The browser-based tool uses AI and deep learning to analyze RGB images of soybean plants and identify eight health conditions: Bacterial Blight, Cercospora Leaf Blight, Downy Mildew, Frogeye Leaf Spot, Healthy, Potassium Deficiency, Soybean Rust and Target Spot.
Farmers can photograph a soybean plant showing unusual symptoms, upload the image to SoyAI and receive a diagnosis indicating one or more potential diseases or disorders. The tool also provides a percentage representing the algorithm's confidence in its prediction.
The technology is designed to give farmers faster access to information about emerging crop health problems. Early identification can help farmers make timely management decisions, potentially reducing chemical applications, lowering production costs and supporting more sustainable crop production.
The initial release is being used for field evaluation, with an Android and iPhone app planned for future release. Feedback from users will help the development team improve the system and guide future enhancements.
A key challenge in developing SoyAI has been the limited availability of large image datasets for individual soybean diseases. The AI system did not have thousands of images representing every disease during its initial training, leaving room for improvement in prediction accuracy.
Field testing will help researchers validate and refine the system. As additional images are submitted, agricultural experts will verify the images and predicted diseases, providing additional information to improve the confidence of future diagnoses.
The next phase of the project will focus on collecting more field data, improving the accuracy of existing disease classifications and adding new soybean disease categories.
Future development could expand SoyAI beyond smartphone-based diagnosis. The lab is exploring integration with drone technology and edge computing, which could allow farmers to detect, identify and monitor soybean diseases remotely and in real time across larger areas.
The University of Maryland team also is developing diagnostic tools for other crops. One project is focused on corn diseases, including tar spot, an emerging challenge for corn farmers.
Source: University of Maryland, "SoyAI Puts Instant Diagnostic Power in Soybean Farmer’s Hands"