AHFML: An Intelligent Aquila Optimized Hybrid Classification Framework for the Detection of Mango Leaf Disease

With an annual production of almost 50 million metric tons, mangoes are one among the highly consumed fruits globally. However, they commonly experience dieback, which is characterized by twig desiccation from the top down, followed by leaf discoloration, drying, and eventually shedding. The study suggests a hybrid classification scheme for mango leaf ailments detection. The model follows preprocessing of images which include rescaling, greyscale conversion and noise removal by applying Weiner as well as Median filters. The images are then optimized with the help of histogram equalization based on the Gaussian Mixture (GM) model. Followed by this, feature extraction is done by a dense Convolutional Neural Network. An improved Aquila Optimized ResNet101 model is employed for the classification. The model-performance is analyzed by comparing with some existing research works. And proposed model found to have 99.2% accuracy, 98.9% sensitivity, 99.1% specificity, 99.1% precision, and 98.9% F1 score rates.