Design of image recognition algorithms for Chinese Herbal medicine plants based on deep learning

EDN: https://elibrary.ru/CIXDJJ

Abstract

Chinese herbal medicine identification is of great significance to the pharmaceutical industry and the sustainable development of herbal production. Traditional identification relies on feature engineering with expert participation, featuring high labor costs and decreasing recognition accuracy as the number of species increases. This paper designs a recognition algorithm based on deep learning, constructs a sample set expanded to 45,236 images through data augmentation, and selects the EfficientNet B2 model based on the TensorFlow and Keras frameworks. With an accuracy of 90.66%, it provides a high-performance solution for Chinese herbal medicine identification and opens up new paths for the application of deep learning in this field.