Sugarcane Plant Disease Detection and Classification using Machine Learning, Deep Learning and Transfer Learning

Abstract

Sugarcane, a vital crop for global sugar and biofuel production, is frequently affected by various diseases, including red rot, smut, and leaf scald, which threaten crop yields and quality. Traditional disease detection methods, such as visual inspections and laboratory testing, are often time-consuming, costly, and impractical for large-scale applications. This research explores the applications of machine learning (ML), deep learning (DL), and transfer learning (TL) in automating sugarcane disease detection. By examining the strengths and limitations of these approaches, this study aims to provide a descriptive analysis of how each technology can enhance disease detection accuracy, efficiency, and scalability. ML offers high interpretability and efficiency for disease detection with moderate data, whereas DL excels in recognizing complex patterns within large datasets. TL, through the adaptation of pre-trained models, proves effective even with limited disease-specific data, making it suitable for agricultural applications. A comparative analysis highlights how hybrid approaches that integrate ML, DL, and TL can address unique challenges in sugarcane disease detection, paving the way for future advancements in scalable, real-time disease management systems. The findings suggest that the integration of these AI technologies holds significant potential to revolutionize agricultural practices, making disease management more accessible and sustainable for sugarcane farmers worldwide.