Efficient segmentation and classification of weed images using deep learning techniques

Abstract

Crop production is the main activity of agriculture, which is newlineresponsible for global food security, nutrition and management. Crops are newlinebeneficial plants that are mass-produced and utilized in the commercial process. newlineIt is essential to look at the new trends, scientific approaches, and accelerators newlinethat can support the agricultural activities. The farmer can reduce their workload newlineby employing technology that improves the crop quality. Weeds are plant newlinespecies that outcompete crops in agricultural fields, reducing crop production newlineand causing severe economic harm to farmers. They increase farming costs and newlineslow irrigation operations by making farm machinery difficult to operate. newlineThe weed destruction process is necessary to save soil nutrients for crops, but newlinethere are less harmful weed species that could be left alive to increase diversity newlineon the farm. Computer vision technology has recently been used to do numerous newlinesmart farming tasks, such as plant species type detection, plant disease newlineclassification, and crop yield prediction. newlineControlling weeds is a crucial part in raising agricultural output. newlineIdentification of weeds and automation of weed removal are two crucial newlineprocesses. However, locating and classifying the weeds in precise level is newlineessential for effective weed removal process. A quick and accurate newlineimplementation of plant detection and classification is required, that categorises newlinethe vegetation into crops, weeds and non-harmful weeds in real-time. The weed newlinemanagement process is intended to be fulfilled by any computer vision newlinetechnologies such as deep learning, and autonomous robot. newlineThe major objective of this research work is to segment and classify newlinethe leaf images using deep learning techniques newline