Enhancing Coffee Production: Classification of Coffee Leaf Diseases Using Optimizer-Driven Spatial Attention CNN and MSA Optimization

Coffee is commercially traded tropical crop globally, constituting the majority of export revenues and serving as a vital economic source for producing nations. Farmers must enhance and oversee coffee production and quality to satisfy global market demands. Coffee leaf disease is a major determinant that diminishes coffee quality besides yield. This research study seeks to precisely classify and identify four principal types of coffee leaf diseases (phoma, miner, rust, and Cercospora) in photos utilising deep learning (DL) constructions, which represent the most advanced artificial intelligence (AI) methodologies. To introduce an optimizer-based Spatial Attention Mechanism Convolutional Neural Network (Op-AM-CNN). To refine the proposed method, this work presents a Mantis Search method (MSA), which has three optimisation phases: prey search (exploration), prey assault cannibalism. A data pre-processing and augmentation technique is employed to improve quality and upsurge the quantity of data samples, hence improving the training of the suggested technique. This study's evaluation indicates that the projected architecture surpassed all existing state-of-the-art deep learning representations, with a validation accuracy of 96.14%.