Tea Leaf Disease Detection Using Deep Learning-based Convolutional Neural Networks
Somya Srivastav, Kalpna Guleria, Shagun Sharma
Abstract
Tea production is a crucial agricultural sector in India that significantly contributes to the country's economy. India is recognized as the world's top tea consumer and the second-largest producer of tea. This industry provides both direct and indirect employment to millions of individuals, especially those residing in rural areas, where it serves as a primary source of income. The growing popularity of deep learning has led to a significant increase in research interest in image recognition technologies, particularly in fields such as computerized image classification and the detection of plant diseases. Each year India produces a huge amount of tea leading to an increase in the economy. However, there is also a vast volume of tea waste because of the tea leaf disease which is required to be resolved to keep the constant industry's success. Diseases affecting tea leaves can impact their growth rate, thus affecting the crop's overall yield. Therefore, this study aims to develop a convolutional neural networks (CNNs) model to diagnose tea plant diseases using a collection of leaf images. CNN is an ideal model for this application as it can effectively analyze both tea leaves and other plant leaves to improve diagnostic accuracy. Additionally, the proposed method has analyzed the tea leaf images which achieved an accuracy rate of 84% along with a very less error rate which is 16% by using the “ADAM” optimizer. Furthermore, in the future, the images can also be enhanced with augmentation methods to achieve better results.