HARNESSING MACHINE LEARNING FOR EFFECTIVE ORANGE FRUIT DISEASE CLASSIFICATION

Abstract

Abstract- A globally grown fruit, oranges are cherished for their health benefits and are often enjoyed by those mindful of their nutrition. In the food industry, proper classification of oranges is essential for sorting, grading, and ensuring product quality. The potential of this technology for disease prediction on farms has yet to be fully exploited. This research aims to address this by classifying four common orange diseases citrus canker, black spot, greening, and scab using pre-trained models like InceptionV3, ResNet50, DenseNet, SqueezeNet, and MobileNet. In addition, it evaluates several optimization strategies such as Root Mean Square Propagation (RMSprop), Adaptive Moment Estimation (Adam), and Stochastic Gradient Descent with Momentum (SGDM). The effectiveness of baseline learning and transfer learning strategies in improving the models' classification performance is also examined in this paper. Our findings show the potential of cutting-edge deep learning methods in agricultural applications by showing notable increases in the detection accuracy of citrus fruit diseases.