Edge Computing Based Early Yellow Rust Disease Detection in Wheat Plants

The agriculture industry contributes most to expanding economies and populations and is essential to the production of high-quality food. Plant diseases are dependent on a variety of environmental variables that can significantly impact the production gains. Manual approaches for identification are laborious and prone-to-error. The advent of computer vision along with the growing trend of cloud computing, have opened the door for deep learning-based plant disease diagnostics. However, a reliable and fast internet connection is required to use cloud computing which is not feasible in most developing countries. In this work, we have suggested a method for detecting wheat crop yellow rust disease using edge computing. Jetson Nano is used to create typical benchmark deep learning models, including ResNet-18, ResNet-50, RegNet_x_3f_2gf, EfficientNet-B2, MobileNet V3, and DenseNet-101. The yellow rust dataset is used to track the training times and detection accuracies of these models. Based on the testing results, the top-performing networks are ResNet-18 and DenseNet-101. Out of the six models under evaluation, they had the least loss of less than 0.30 with the accuracy of 87%. In comparison to other deep learning models, Resnet-18 needed the least amount of training and testing time, making it the best model for edge computing.