Detection and Classification of Various Diseases in Cotton Crops using advanced Neural Network Approaches

The cotton crop is important in India's agricultural terrain, constituting an important business experience. The vulnerability of this crop to various diseases is intricately linked to environmental changes and soil grade. Early detection of these diseases is crucial, followed by applying appropriate pesticides to reduce their impact on crop yield. Our research proposes a model utilizing Neural Networks to identify five distinct cotton crop diseases. We created a dataset containing images representing four specific disease classes: Aphids, Bacterial Blight, Curly Leaves, Powdery Mildew, and Verticillium Wilt. We included a class for images of healthy leaves. Our synthetic dataset contains 1200 images of cotton plant leaves. The model implementation involved the use of MobileNet, VGG-16, VGG-19, and custom CNN models for image classification. Among all the models, VGG-19 achieved the highest accuracy at 97.08%.