Deep Learning Architectures Performance in Plant Leaf Diseases

Abstract

Developing artificial intelligence applications continue to make our lives easier. Imageprocessing technology has been developed for field-useful studies in medical science, education, finance,agriculture, industry, security, and many other sectors. For agricultural products, good works are done withartificial intelligence to detect plant diseases and take precautions accordingly. In our study, a comparisonwas made with deep learning methods on the images of the sugar cane plant in different categories. TheVGG-19 architecture, which was classified separately from the 5 pre-trained architectures AlexNet,DarkNet-53, GoogLeNet, ResNet-50, and VGG-19, reached the highest accuracy with 92.2%.

Plant phenotyping relevance

サトウキビ葉の画像から病害を分類する深層学習手法を比較・評価しており、植物の病害状態を画像ベースで推定する方法が研究の中心です。

Code and data availability

The paper's phenotyping input is a publicly available sugarcane leaf disease image dataset (2,569 images; Healthy, Mosaic, Redrot, Rust, Jaundice) deposited on Mendeley Data by the cited authors (Daphal & Koli, 2022, doi:10.17632/9424skmnrk.1). No author analysis code, trained models, or supplementary assets are stated