Images and annotations of Puccinia sorghi infected maize leaves

The data within this repository contains all of the training and testing data used to train and evaluate 70 U-Net neural network models that were developed to quantify common rust symptoms, as caused by the rust fungus Puccinia sorghi, on maize leaves. This data was originally used to test the effect of the types of training data used, increasing amounts of training data, and two threshold strategies.

Each image has a corresponding annotation file that contains the region coordinates for common rust pustules. The evaluation files contain the neural network model output for a given training strategy. As each model was tested at two thresholds, there are 140 total evaluation files.

Commercial licensing enquiries should be directed to the Iowa State University Office of Innovation Commercialization at: licensing@iastate.edu, or 515-294-4740