OHMCPD: Optimized Hybrid Deep Learning Model for Classification of Cotton Plant Diseases

The quantity and quality of crop output heavily influences the economic prosperity of the country. Leaf disease detection early on can boost yield and profit. Many image processing methods have been developed to detect leaf diseases. Farmers may find general observations to be time-consuming, expensive, and sometimes unreliable. The image is scaled, converted to grayscale, and denoised by a combination of various Dividend and Weiner filters before undergoing further processing. Another technique for improving images is histogram equalization based on Gaussian mixtures. A thick convolutional neural network is used to assist in image feature recovery. The improved ResNet101 model adapted to Aquila is used here for classification. This model is used to evaluate the accuracy of cotton leaf pest and disease classification. With an accuracy of 99.09%, sensitivity of 97.84%, specificity of 98.10%, precision of 97.92%, and F1 score of 98.68%, the suggested model performs well.