AI-based smart agriculture 4.0 system for plant diseases detection in Tunisia

Plant diseases pose a significant problem for agricultural sustainability, notably reducing crop quality and yield. Addressing this challenge, this study introduces an AI-based smart agriculture 4.0 system specifically designed for diagnosing plant leaf diseases. Using a self-generated dataset collected from the Monastir region of Tunisia, the system employs various convolutional neural network (CNN) architectures, including AlexNet, VGG16, VGG19, ResNet50, ResNet152, and DenseNet121, to identify diseases in tomato and pepper plants. A comparative analysis of these CNN models was conducted, highlighting the efficacy of each in disease identification. Notably, the study reveals that a transfer learning-based ResNet152 model achieved the highest accuracy rate of 99%, outperforming other models. Hence, the proposed approach is promising and has the potential to make a significant impact on effective disease detection.