Comparative Analysis of Deep Learning Architectures for Plant Leaf Disease Detection

Abstract

In recent years, significant progress has been made in effectively using deep learning techniques for the diagnosis and classification of foliar diseases in plants. This study compares the performance of five different deep learning models - VGG16, InceptionV3, AlexNet, MobileNetV2, and a specially developed CNN architecture - in terms of their ability to diagnose and classify foliar diseases in plants. The dataset was subjected to horizontal and vertical rotations to enhance the models' generalisation ability, and data augmentation methods such as brightness adjustments and noise addition were employed. A total of 13,716 images were analysed, split into 70% for training, 20% for validation, and 10% for testing. According to the results, the InceptionV3 model outperformed the others in terms of performance measures such as accuracy, loss, precision, sensitivity, and the F1 score. This study highlights the importance of deep learning approaches in the early and accurate diagnosis of plant diseases, revealing that these methods can help farmers and agricultural experts minimise production losses and promote healthy plant growth. The comparative analysis provides valuable insights for developing intelligent, customised agricultural surveillance systems.