Exploring DenseNet121 for Disease-Specific Classification in Orange Crop Health Monitoring

This paper analyzes four forms of orange sickness categorization using the DenseNet121 model: blackspot, canker, greening, and healthy orange. Using an annotated photo dataset, the model exhibited high reliability on all performance metrics. For all classes, accuracy dropped between 0.94 and 0.96; recall fell between 0.94 and 0.97; and F1-scores fell between 0.94 and 0.96. With a continually high classification accuracy of 96.14%, the model showed to be particularly good at correctly recognizing diseases. These data demonstrate that DenseNet121 is appropriate for the detection of orange disorders, hence it is a helpful tool for early identification and disease management in agricultural approaches. This work is crucial as it will assist to minimize losses by means of accurate and early illness identification, hence boosting agricultural productivity. This facilitates enhanced resource management and more ecologically friendly agriculture approaches including concentrated pesticide use. Moreover, the technology gives scalability and enables one to adapt the approach to different crops and ailments. Future research will focus on including real-time deployment systems—like mobile apps—to help farmers access disease diagnoses anywhere. This approach seeks to bridge the gap between sophisticated machine learning solutions and practical field applications in precision agriculture.