A Banana Leaf Disease Detection System of an Energy-Efficient CNN-Based Framework

Diseases that afflict banana leaves have a substantial influence on the productivity and quality of crops, representing an important obstacle for sustainable agricultural systems and food security. Rapid and precise disease identification is crucial for sound crop management and loss minimization. The major benefit associated with leveraging machine learning and deep learning technologies within the agricultural domain lies in their capacity to facilitate fully automated diagnosis of plant disease through image-based analysis, particularly effectively complemented by detection certainty under smart agriculture systems. The paper proposes an efficient CNN-based automated framework for banana leaf disease detection. A thorough review of literature, insight into existing knowledge gaps, and a well-defined research hypothesis are elements that create a solid foundation for research. The framework here integrates preprocessing, data augmentation, and a transfer learning–based EfficientNet architecture with an emphasis on computational efficiency that is suitable for real-time deployment. This paper showcases the initial phase of a multi-phase research study aimed at conceptual design and methodological development.