Detection and Recognition of Traditional Chinese Medicine Slice Based on YOLOv8

Abstract

Traditional Chinese medicine (TCM) herbal pieces are a type of TCM preparation with complex production processes and a wide variety of species. It is a challenging task for ordinary people to identify TCM herbal pieces. To solve this problem, this paper proposes an object detection method for TCM herbal slice based on YOLOv8. The YOLOv8 model is trained on a Chinese herbal slice (CHS) dataset that is collected and labeled from the internet, and with the Mosaic data augmentation, the average precision of YOLOv8 model on all categories can reach 95.8%, with a model size of 5.94MB, and can achieve <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$30\sim 45$</tex> FPS on the MX150 graphics card environment. Compared with image classification mode, the object detection mode of YOLOv8 can better perform efficient and accurate detection and classification of CHS, which can effectively improve the recognition rate and popularization of CHS, and enable more people to benefit from the culture of TCM.

Topics & Concepts