Enhanced cotton leaf disease classification Through ConvNeXt and Random Walk Grey Wolf Optimization

India has the highest number of cotton-cultivated land globally, thus placing it among the biggest producers of cotton in the world. Despite the huge cotton production, cotton leaf diseases still remain one of the most significant causes of yield losses in the crop. Accurate and early diagnosis is one of the first steps towards finding the solutions to mitigate the effects of the disease. In this research paper, we present the ConvNeXt deep learning framework used along with Random Walk Grey Wolf Optimizer (RWGWO) to diagnose and classify various diseases in cotton leaf. In its performance, ConvNeXt shows better accuracy when compared to the traditional CNN, Self-Organizing Map Artificial Neural Networks (SOMANN), Support Vector Machine (SVM) and other hybrid models with classification accuracy of 98.39%. Among others, our model successfully diagnoses and classifies the diseases of bacterial blight and aphid infestation, powdery mildew, armyworm, target spot, and healthy cotton leaves. All this is possible through a carefully chosen data set with proper data augmentation. The results of this research indicate that there is a high potential in using deep learning technology to improve agribusiness and textile fibers supply chain through early detection and management of the diseases outbreaks, which reduces the waste of pesticides associated with the late disease diagnosis. On the other hand, it should be noted that the use of ConvNeXt has limitations since larger models of this framework (ConvNeXt-Large and ConvNeXt-XL) have high computational complexity and energy consumption. The future work will include using field data sets and implementation of multi-modal learning, which will take into consideration various environmental factors.