RoboCrop: Automated Classification of Soybean Diseases

*Experimental tool. Use at own risk.*

Classifications are based on a convolutional neural network, trained on digital images from Alabama soybean fields in 2020/2021.

Classifications work best on horizontal color photographs of single leaves, either attached to the plant, or placed in grass or on a flat white surface. Examples of ideal photos are displayed at the top of the page.

Currently, the model is trained to recognize healthy plants, along with the following diseases:

Bacterial Blight, Cercospora Leaf Blight, Downey Mildew, Frogeye, Potassium Deficiency, Soybean Rust, Target Spot

*Disclaimer: Many diseases have similar visual patterns. If a diseased image category is uploaded that the model has not been trained on, it will likely still predict, rather confidently, one of the diseases that it has been trained on. This is due to high pattern similarity across all diseases. In any and all cases, further resources (see below) should be used to confirm disease identification. *

Alabama Cooperative Extension System (ACES)

ACES: Soybean IPM Guide

Crop Protection Network: Soybean Encyclopedia