Harvesting Insights: Revolutionizing Orange Disease Detection with DenseNet-121 for Sustainable Farming

In this work is examined the application of DenseNet-121, a deep learning architecture, for disease detection in orange crops-more especially, blackspot, canker, greening, and healthy conditions. Major challenges in orange cultivation, various diseases-including blackspot caused by Phyllosticta citricarpa and greening (Huanglongbing) brought on by Candidatus Liberibacter severely influence yield and fruit quality. Automated solutions are needed since conventional manual illness detection techniques are time-consuming and frequently erroneous. Particularly with limited datasets, DenseNet-121 is well-suited for picture classification tasks in agriculture since its densely connected layers improve feature reuse and efficiency. This work trained and validated the model using a $\mathbf{1, 1 6 4}$ picture dataset including image augmentation methods to increase robustness. Transfer learning helped to improve the pre-trained weights of DenseNet-121 in line with the specific aim of classifying four orange scenarios. The model proved helpful in exactly identifying and detecting diseases with an amazing accuracy of 97.4% over 100 epochs. This paper emphasizes the possibilities of deep learning techniques like DenseNet-121 to automate disease identification in oranges, therefore lowering dependency on human inspection and facilitating more reasonably priced, efficient agricultural ways. Their support of real-time monitoring systems and the spread of comparable technologies to other crops helps to support food security and environmental sustainability by means of which the results have major impact on sustainable agriculture.