Leaf Disease Detection Using Transfer Learning

Abstract

The early diagnosis of leaf diseases is essential for maintaining the health and yield of crops. With advancements in deep learning and computer vision, transfer learning has emerged as a powerful technique for solving complex image classification problems. This paper presents a comparative analysis of three widely used convolutional neural network (CNN) models, namely ResNet, MobileNet, and VGG16, for leaf disease detection using transfer learning. The experimental results are evaluated according to a number of performance indicators, including as accuracy, precision, recall, and F1-score. Results show a test accuracy of 89.75%, 88.05% and 92.73%, respectively. The comparative analysis of the models provides insights into their respective strengths and weaknesses. Furthermore, visualizations of the confusion matrix and sample predictions offer a comprehensive understanding of their classification abilities. The paper also examines the practical implications of the models for real-time leaf disease detection. Factors such as inference time and computational resource requirements are considered to assess their suitability for deployment in real-world scenarios. The analysis tries to direct practitioners in choosing the best model for their particular application, taking the trade-off between accuracy and efficiency into consideration. In conclusion, this study provides a detailed comparative analysis of ResNet, MobileNet, and VGG16 models for leaf disease detection using transfer learning. The findings shed light on the performance, efficiency, and practical implications of these models, facilitating informed decision-making for researchers and practitioners working in the field of agricultural plant disease detection. The results also suggest potential avenues for future research and improvement in this domain.