TOMATO LEAF DISEASE DETECTION USING YOLOV9 AND COMPUTER VISION

Abstract

Tomato production is very significant in agriculture and food security and it implies that the health of these plants is very vital in this production. Thus, diagnosing the diseases of tomato leaves at the initial stage and with high accuracy allows for minimizing the losses of the crop yield and improving the quality of the tomatoes. In this paper, research demonstrate the YOLOv9 model for detecting several diseases in tomato leaves. The dataset for this work consists of various diseases such as bacterial spot, early blight, late blight, leaf mold, target spot, and black spot. For the performance of the YOLOv9 model, mAP, precision, recall values were used. From the results obtained it is clear that YOLOv9 obtains a mAP50 of 0.466 and mAP50-95 of 0.305. Most striking is the model’s performance in classifying healthy leaves, with a precision of 0.659 and a recall of 1. In the accuracy assessment of the relative program, the recall achieved the highest value of 0 in early blight detection 0.854 and a mAP50 of 0.675. However, the general detection accuracy of the model was low in some of the diseases such as leaf mold and target spot and this could be due to many factors. The research also lays focus on the applicability of YOLOv9 in the automated systems for disease diagnosis and further emphasizes on the fact that the detection  algorithm needs to be optimized for less common diseases. Thus, AI model as YOLOv9 in agriculture can be beneficial for crop health and decrease the level of dependence on chemical treatments for sustainable farming practices. Hence, this research avails a comprehensive diagnostic tool for early detection of diseases that address the need of the society in developing innovative precision agriculture technologies that realize high production of crops and food sufficiency.