An Ensemble Learning Framework for Predicting Tea Leaf Diseases Using Stacking Classifiers
Main Article Content
Abstract
Diseases in tea crop form one of the greatest obstacles in farming that can result in a decrease in yield and economic profit. The paper introduces a machine learning based technique of early detection and classification of tea crop diseases using stacking ensemble technique. The system envisioned will combine various information sources such as data on environmental sensors (temperature, humidity and soil moisture), weather and leaf images. Upon preprocessing and feature extraction, a wide range of base classifiers, including: Random Forest, Support Vector, Xgboost, K-Nearest Neighbor, and Convolutional Neural Networks are trained to obtain various patterns of disease symptoms. These base models are then aggregated to form the final classification by a meta-learner and the performance of the model is measured by using K-fold cross validation and various measures of performance such as accuracy, precision, recall, and F1-score. The results of the experiment prove that stacking ensemble works better when compared to single classifiers, and it gives stable performance with unbalanced and noisy data. The methodology provides the solution to smart agriculture in a scalable, real time method and contributes to proactive disease control in tea plantations.