Early Detection and Identification of Blister Blight Disease in Tea Plantations Using Deep Learning and Real-Time Environmental Data

Abstract

Tea, one of the world's most widely consumed beverages, faces significant threats from Blister Blight disease, particularly in hot and humid regions. This disease, caused by the fungus Exobasidium vexans, leads to substantial crop losses, negatively impacting both the quality and quantity of tea production. This paper presents an integrated approach to tackling Blister Blight disease by combining Machine Learning techniques, and real-time environmental data analysis. Utilizing image data from various sources, pre-processed for deep learning models, this approach aims to accurately detect and classify Blister Blight infected leaves. Additionally, real-time environmental data analysis is used to predict Blister Blight onset by identifying critical weather thresholds and correlating them with disease occurrence. The implementation of an automated alert system delivers timely warnings based on environmental conditions, enabling proactive interventions to mitigate Blister Blight’s impact on tea production. This approach not only improves the accuracy and efficiency of disease detection but also aligns with the broader trend of incorporating technology into agriculture for sustainable and efficient practices. The proposed framework represents a significant advancement in agricultural disease management, offering a scalable and adaptable solution that can be applied to various plant diseases. Through this work, we aim to contribute to the long-term sustainability of the tea industry, ensuring economic stability for regions dependent on tea cultivation and promoting the broader adoption of precision agriculture technologies.