Implementation and Evaluation of Various Deep Learning Models for the Diagnosis of Cotton Leaf Diseases

Cotton, a vital cash crop, suffers significant yield losses due to various leaf diseases. Early and accurate detection of the diseases is crucial for implementing control measures. This research explores the potential of Convolutional Neural Networks (CNNs) for cotton leaf disease classification. We have evaluated five pre-trained CNN architectures: DenseNet, MobileNet, ResNet, VGGNet, and GoogleNet. A publicly available cotton leaf disease dataset is used for training and testing. GoogleNet achieved the highest accuracy (95%). Detecting diseases in cotton leaves involves various key steps, first comes image preprocessing followed by segmentation then feature extraction and finally classification. The accuracy of these steps profoundly affects the reliability of the disease classification algorithms. Recent studies have assessed cutting-edge techniques in this area, showing promise for improving the crop yield and reducing losses as much as possible. Early disease detection allows timely interventions to control spread which in turn results in preventing any significant damage. This research compares the latest advancements, evaluating their practicality in the real time scenarios.