Research on the Classification and Recognition of TCM Based on YOLOv8

Chinese medicinal materials are diverse in types and complex in morphology. Traditional manual identification methods are inefficient and highly subjective, making it difficult to meet the requirements of modern Chinese medicine industry for identification efficiency and accuracy. To address this issue, this paper constructed its own dataset of Chinese medicinal material images and trained and evaluated it based on the YOLOv8l object detection algorithm. The experimental results show that the Top-5 curve of the YOLOv8l model is stable in the range of 0.8–0.9 and significantly higher than the Top-1. Although there is a certain degree of overfitting in the training set, the model has a strong ability to distinguish similar medicinal materials, thus having practical promotion value. This study conducted systematic design from aspects such as model selection, dataset construction, and parameter configuration, providing an efficient and practical solution for the intelligent identification of Chinese medicinal material images.