Abstract
Betel leaf cultivation holds a significant role in a country’s agricultural economy. Betel leaf cultivation holds significant cultural, medicinal, and economic importance in several countries. However, its growth, quality, and market value are frequently threatened by a wide range of fungal, bacterial, and viral diseases. However, the growth and quality of these leaves often suffer due to various diseases, posing significant challenges for accurate and timely manual identification. Addressing this, a novel deep learning-based framework, named BetelProNet, is proposed to automate early disease detection. BetelProNet leverages the MobileNetV3 architecture, enhanced with task-specific fine-tuning and efficient depth wise separable convolutions, to achieve superior accuracy while maintaining computational efficiency. Extensive training on diverse datasets demonstrates the model’s adaptability across varied data sources. The model showcased outstanding performance, It achieves exceptional results, including 98% accuracy, 99% precision, 96% recall, and a 97% F1-score across disease categories. Additionally, it showcases superior inference speed compared to existing models. A comprehensive comparative analysis underscores the BetelProNet framework’s superiority over various pretrained models as well as traditional machine learning and deep learning techniques. This model emerges as a valuable tool for rapid, reliable disease identification in betel leaves, contributing to better crop management and enhanced agricultural output.
Keywords: Betelpronet, Deep Learning, Disease Detection, Mobilenetv3, Precision Agriculture.