Abstract

Tea, also sometimes referred as Chai is one of the popular aromatic beverages consumed across the world, and it is consumed in various forms & flavours. Tea plants are generally cultivated in tropical and subtropical climatic conditions. China, India, Kenya & Sri Lanka together contributes to more than 75% of the world tea production, with India ranks only next to China with 1.2 million tonnes of tea production in 2020 [10]. Timely identification of the presence of disease in tea leaves is a crucial task during the cultivation processes & prevent rapid spread of the same. The manual identification of the disease is more linked to the skill set, knowledge & expertise of the particular individual. Image processing technique through Convolution Neural Network (CNN) is being utilized in this paper for Disease identification & classification with prime focus on Tea leaves. The objective is to have a automated system for classification of diseases present in the tea leaves images with the help of dataset, and produce the results along with its cause, symptoms & precautions. This non-manual oriented & non-destructive disease identification & classification system is expected to be of high assistance to the tea cultivation & processing industries to have improved monitoring process. This deep-learning model uses CNN technique for feature extraction and classification processes and reached the overall accuracy of 88% (training accuracy 92% and validation accuracy 80%). The output of the system is shown through Streamlit framework for a user-friendly illustration.

Authors

G. Sekar¹, M. Najma², P. Manoj³, M. Yuvaraju⁴
V.S.B. College of Engineering Technical Campus, India^1,3, Assistanz Networks Pvt. Ltd, India², Anna University, Regional Campus, Coimbatore, India⁴

Keywords

CNN, Deep-learning, Feature Extraction, Stream Lit, Tea leaves Disease