Detection and Classification of Cotton Plant Disease Using Deep Learning Network

Abstract-This research aims to address critical challenges in agricultural sustainability by proposing a multifaceted approach to the detection and prediction of diseases affecting cotton plants. The objectives of this study are threefold. Firstly, the research focuses on the classification of cotton plant leaves, essential for accurate disease diagnosis. Through dataset analysis, normalization techniques, and feature extraction using Local Binary Patterns (LBP), cotton plant leaves are effectively differentiated from other foliage. Classification is accomplished utilizing Lightweight Convolutional Neural Networks (CNN), with performance parameters rigorously evaluated to ensure efficacy. Secondly, the study extends its scope to the classification of diseases affecting tomato plant leaves, offering insights into disease identification methodologies applicable to cotton plants. Leveraging the Coral Reef Optimization approach for feature extraction and a hybrid classifier comprising ResNet50 and VGG16 architectures, the system achieves precise disease classification. Lastly, the research addresses the critical need for predictive analytics in disease management by forecasting the occurrence of diseases in cotton plants. Utilizing historical time series weather data, machine learning and deep learning models, specifically Quantile Regression Forests coupled with Long Short-Term Memory (LSTM) algorithms, predict temperature and relative humidity parameters crucial for disease occurrence. By integrating these objectives, this study endeavors to provide a comprehensive framework for proactive disease management in cotton cultivation, thereby contributing to sustainable agricultural practices and food security.