Abstract: The “Intelligent Detection of Sapthashira and Its Diseases” project develops a deep learning framework for the automated identification of diseases affecting Sapthashira (betel) leaves, even when obscured by overlying pepper leaves, while providing targeted preventive recommendations for each diagnosed condition. Leveraging an EfficientNetB4 transfer learning architecture implemented in TensorFlow-Keras, the system preprocesses and augments input leaf images to achieve robust classification of healthy versus diseased specimens across diverse real-world scenarios. Integrated into a Flask-based web application, it enables users to upload images for real-time diagnostic output, including evidence-based interventions such as isolating infected plants, excising severely compromised foliage, and administering specified fungicides or bactericides—thereby optimizing crop protection, minimizing yield losses, and reducing superfluous agrochemical applications. This extensible platform establishes a scalable foundation for precision agriculture, with potential for adaptation to additional pathogens and crop varieties in subsequent developments.

Keywords: deep learning, EfficientNetB4, Sapthashira diseases, transfer learning, Flask web application, and preventive measures.

Downloads: | DOI: 10.17148/IARJSET.2026.13140