Efficient plant leaf disease detection using optimization enhanced deep learning framework

Abstract

Agriculture plays a pivotal role in driving the economy of newlinenumerous countries worldwide and is the cornerstone of the livelihoods of a newlinemajority of people. Despite the rising demand for vegetables and allied newlineagricultural products, farmers are unable to meet the demand, owing to low newlineproductivity. Chiefly, the agricultural sector is concerned with maximizing newlineproductivity, quantitatively and qualitatively, in the face of adverse natural newlineconditions. Given that plant disease is a natural occurrence, its early detection newlineboosts both production quality and volume. A visual plant disease diagnosis is newlinea time-consuming process, inaccurate, and only practicable across a few newlinelocations. However, automated detection methods call for reduced time and newlinework while simultaneously resulting in greater accuracy. newlineAn efficient plant leaf disease identification system is needed to newlineidentify invasive pests and diseases at the outset. Most recently, techniques of newlinemachine and deep learning have been applied to recognize and diagnose plant newlineleaf diseases. Deep learning methods execute training without supervision newlinedirectly from the source image to acquire information about image functions newlineat multiple levels, including semantic characteristics at the low, middle, and newlinehigh levels. Existing deep learning classification and segmentation models newlinemay be enhanced through optimization, feature selection, and feature newlinecombination techniques, resulting in far more accurate disease prediction and newlinemask generation. newline