Leaf Disease Detection in Plant Care using CNN Architecture: AlexNet and ResNet-50 Models

Abstract

Global agricultural productivity is integral to fulfilling basic nutrition needs and economic growth. Moreover, plants are essential in protecting the environment and food chain balance. However, plant growth and health naturally depend on whether they are affected by various diseases. Agricultural cultivators in remote regions often lack precise information on effective disease detection methods, leading to significant crop losses. Manual observation is an unreliable technique for disease detection, making it challenging to identify and address issues promptly. Accurate disease detection through analyzing leaf images can be a crucial tool for quickly and easily noticing and solving potential issues in digital cultivating. This paper proposes a method for disease detection in plant care using the image dataset of tomatoes and potatoes to help cultivators better manage plant health. The approach leverages image analysis of plant leaves, employing AlexNet and ResNet-50, two well-known convolutional neural network models. This approach has utilized a dataset from Kaggle that includes images of tomato and potato plant leaves to explore leaf diseases. Hence, to detect leaf disease early, processes have been performed that involve preparing images, augmenting them, identifying important features, and classifying them through AlexNet and ResNet-50, including model evaluation using accuracy as the metric. According to experimental results, the proposed work achieves an overall 95.9% accuracy of the AlexNet and 97.3% accuracy of the ResNet-50 for identifying leaf diseases. It contributes to agriculture by providing an effective method for detecting plant leaf diseases and taking timely preventive measures for plant health.