Automated Paddy Leaf Disease Identification using Visual Leaf Images based on Nine Pre-trained Models Approach

Abstract

The recent rise in paddy leaf diseases poses significant challenges, emphasizing the need for focused research and rapid implementation of an artificial intelligence technique for the identification of crop leaf disease. As a staple food for over half the global population and a key ingredient in global cuisines, paddy offers numerous health benefits but is hindered by diseases like brown spot and blast disease. Effective paddy leaf disease management requires precise classification. This study used a public dataset and Artificial Intelligence to identify and classify these diseases. We applied a deep Convolutional Neural Network (CNN) and nine transfer learning models (VGG19, VGG16, DenseNet121, MobileNetV2, DenseNet169, DenseNet201, InceptionV3, ResNet152V2, and NASNetMobile) using TensorFlow. Each model’s performance was assessed to find the most effective classification system, covering four disease categories and one non-disease category. The research aimed for shorter training times, higher accuracy, and easier retraining, with DenseNet121 achieving the highest classification accuracy of 97.6% on the paddy leaf image dataset.