Machine Learning-based classification of cotton diseases using mobilenet and Support Vector Machine

Plant disease detection and recognition is one of the most essential components of agriculture and ecosystems. Cotton is an important plant in Egypt, Africa, the United States, China, Australia, India, and many other countries around the world. Therefore, cotton diseases are considered a big problem for the industries and economies of countries. The annual percentage yield is reduced easily. To address this problem, many studies have proposed different methodologies for detecting cotton leaf diseases. In this study, we aim to reduce yield losses caused by cotton diseases and help farmers easily detect cotton diseases before they spread to crops. So, a high-performance model is developed for cotton disease detection from leaves. The study methodology is based on using various techniques for augmenting the data and fine-tuning Support Vector Machine (SVM) hyper parameters to enhance its classification performance. To improve disease detection accuracy, the study also incorporates cutting-edge deep learning models, specifically VGG16 and MobileNet. The results of the study show that the SVM-VGG16 hybrid achieved a commendable 91% accuracy, while the MobileNet model achieved an impressive 99% accuracy. This detailed analysis not only highlights the impact of data augmentation and SVM optimization but also emphasizes the efficacy of deep learning models in advancing the accuracy of cotton leaf disease detection. These findings could have significant implications for precision agriculture and crop health in monitoring systems.