Lemon leaf chemical feature extraction over categorical images using Neural Network

Abstract

This paper convolutional neural networks (CNNs) have shown great promise for the categorization and chemical feature extraction of lemon leaves, among other agricultural applications. This paper investigates the use of CNNs, namely MobileNet, which achieves a maximum accuracy of 84%, for the precise age classification and chemical composition prediction of lemon leaves. In order to discern between the oldest, middle-aged, and young age groups and to forecast nutrient levels like potassium, calcium, magnesium, phosphate, and sulfate, the research uses deep learning algorithms to analyze leaf photos and extract complex information. Preprocessing leaf pictures, training MobileNet on an extensive dataset, and assessing model performance using measures like as accuracy, precision, recall, and F1-score are all part of the technique. The outcomes show how well MobileNet works to achieve high accuracy in tasks involving both chemical feature prediction and classification. By giving farmers strong tools to track leaf health, improve nutrient management plans, and increase crop output in a sustainable way, this work advances precision agriculture. To further improve the scalability and usefulness of CNN-based techniques in agricultural contexts, future research areas include expanding the dataset, investigating ensemble learning strategies, and incorporating real-time applications.