The Automatic Tea Leaf Classification Using Deep Learning Models

Abstract

Abstract: Tea leaf classification plays a crucial role in quality control and grading within the tea industry. However, the manual classification process is time-consuming and subjective. This study aims to address these challenges by developing an automatic leaf quality detection technique. Three deep learning models, namely convolutional neural networks (CNN), Inception v3, and EfficientNet B0, were employed to classify tea leaves based on their visual features. A dataset of tea leaf images was collected and split into training, validation, and testing sets. The models were trained on the training set, with hyperparameter tuning performed using the validation set. The performance of the models was evaluated on the testing set. The results revealed that both convolutional neural networks (CNN) and EfficientNet B0 outperformed Inception v3, achieving accuracy rates of 96.9% and 97.4%, respectively, compared to Inception v3's accuracy rate of 95.9%. These findings demonstrate the effectiveness of deep learning algorithms in tea leaf classification and suggest their potential for further improvement and application in the tea industry.