Leveraging MobileNetV2 for Accurate Detection and Classification of Orange Diseases: A Study on Blackspot, Canker, and Greening Citrus

In this article we examine three main diseases impacting oranges: citrus canker, black spot, and greening citrus. These diseases cause major losses in the citrus industry since they adversely influence the quality and output of orange harvests. Early diagnosis of these diseases is crucial to stop their spread and guarantee the health of the fruit, which is vital for both farmers and the agricultural market, so supporting food security. We addressed this issue using a classification model based on MobileNetV2 architecture with transfer learning. The collection consists in 1164 images divided into four categories: greening citrus, black spot, citrus canker, and healthy oranges. We budget eighty percent for training and twenty percent for testing. Trained and tuned specifically for this task, the MobileNetV2 model generates efficient feature extraction and classification. With a great accuracy of 95%, our model proved effective in separating healthy oranges from those affected by particular diseases. This outcome emphasizes the possibilities of employing transfer learning models for real-time, automated disease detection in orange crops, so providing a useful way to protect crop production and quality and to support sustainable agriculture, economic development, and sensible consumption patterns.