Automated Disease Identification in Mango Leaves Using Image Analysis

Mangoes are vital crop globally, and their economic value necessitates effective disease detection strategies to maintain optimal health and profitability. However, the automatic classification of mango leaf diseases presents significant challenges due to the diverse and often overlapping symptoms exhibited by various diseases. This study proposes about Convolutional Neural Network (CNN) model for classifying diseased mango leaves, utilizing preprocessing techniques to analyze the properties of individual pixels. The CNN model demonstrated superior segmentation accuracy when compared to advanced models such as VGG16, ResNet50, DenseNet, EfficientNet, InceptionV3, and MobileNet, achieving remarkable performance in training accuracy. ResNet50 achieved an astonishing 100 % validation accuracy, highlighting its potential for robust leaf disease detection. The proposed model enhances recognition accuracy by focusing on distinct features that aid in disease diagnosis. An algorithm was also developed to assess mango leaf health via a web application that captures and analyzes leaf images. Despite the extensive variety of mango leaf diseases, the dataset used in this study was categorized into two classes: healthy and diseased leaves. The primary objective of this research is to accurately detect and classify healthy and diseased mango leaves, contributing to early disease detection and effective crop management strategies.