Embedded AI for Wheat Yellow Rust Infection Type Classification

Wheat is the most important and dominating crop in Pakistan terms of production acreage, which grown on 37% cultivated area, accounting for 70% total production. However, wheat yield highly affected by stripe rust, considered devastating fungal disease, causing 5.5 million tonnes loss per year globally. In order to minimize this loss, accurate timely detection rust disease crucial instead manual inspection. Towards end, we propose a system detect classify its infection types into four classes, including healthy, resistant, moderate (moderately resistant moderately susceptible), susceptible. The dataset collected indigenously from National Agricultural Research Centre, Islamabad. A pre-trained U 2 Net model used remove background extract leaf containing disease. Subsequently, two deep learning classifiers, Xception ResNet-50 are applied severity levels, where outperformed with highest accuracy 96%. This research presents comparison between state-of-the-art classifiers accuracy, memory utilization, prediction time, will assist community selecting appropriate plant detection. Moreover, assess external validity, performance these compared existing technique using publicly available dataset, confirms validity results. Additionally, an intelligent edge computing device has been developed, trained deployed, facilitates farmers monitor attack. proposed aimed agricultural employ preventive measures site-specific manner based diagnosis & severity, intended improve quality as well