An Image Based Classification and Prediction of Diseases on Cotton Leaves Using Deep Learning Techniques

Cotton is one of the most important crops but its cultivation is hampered by diseases and infections. Rapid disease identification is the challenge due to lack of infrastructure. This necessities the development of automated methods to predict diseases. So in this research Deep Learning Models (DLM) could be used for the development of intelligent agriculture to effectively determine the spot of the diseased cotton leaf part in the terrain. Semantic image segmentation is considered a major problem in computer vision. So we investigate that deep learning of Convolution Neural Networks (CNN) for the deduction of cotton leaf disease and implement the knowledge of deep learning models and transferring the knowledge to our dataset. In this attempt we employed Image recognition model such as VGG16, Inception-V3, Image data generator, flattening, batch normalization, conv2d and to extract diseased parts in cotton leaf using the dataset of both healthy and diseased leaves. Optimization is used to choose optimal values and image augmentation is done to improve the models. The result demonstrates that all trained models have an accuracy more than 90% in classify cotton leaf disease. In particular VGG16, Inception-V3 attained an accuracy of more than 96%.