Scholarlink

This study explores the application of deep learning techniques, particularly the YOLOv8 model, in the field of herbal medicine identification. Traditional identification methods suffer from subjectivity and inefficiency. In this paper, we propose an improved YOLOv8-based image recognition method for herbal medicine, aiming to enhance accuracy and efficiency. A large-scale image dataset containing various common herbal medicines was constructed, and targeted optimizations were applied to the YOLOv8 model. Comparative experiments with Faster R-CNN, SSD, and YOLOv5 demonstrate that the optimized YOLOv8 model performs exceptionally well in herbal medicine identification tasks, achieving an average accuracy of 95.8%, which is 3.2% higher than YOLOv5 and significantly surpasses other models. In terms of speed, YOLOv8 can process approximately 45 images per second, nearly three times faster than Faster R-CNN.The innovation of this study lies in the application and optimization of YOLOv8 for herbal medicine identification, demonstrating its advantages in accuracy, speed, and practicality. The research outcomes provide a new technical approach for quality control in herbal medicine, potentially advancing the modernization and standardization of Traditional Chinese Medicine.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

The Research on Traditional Chinese Medicine Identification Method Based on YOLOv8 Deep Learning Model

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Rina Wu,
  • Liyang Shan,
  • Huanyu Liu,
  • Jieqiong Li,
  • Yang Liu,
  • Xiurong Li,
  • Xiaowan Li,
  • Jintong Zhang,
  • Siying Tao

摘要

This study explores the application of deep learning techniques, particularly the YOLOv8 model, in the field of herbal medicine identification. Traditional identification methods suffer from subjectivity and inefficiency. In this paper, we propose an improved YOLOv8-based image recognition method for herbal medicine, aiming to enhance accuracy and efficiency. A large-scale image dataset containing various common herbal medicines was constructed, and targeted optimizations were applied to the YOLOv8 model. Comparative experiments with Faster R-CNN, SSD, and YOLOv5 demonstrate that the optimized YOLOv8 model performs exceptionally well in herbal medicine identification tasks, achieving an average accuracy of 95.8%, which is 3.2% higher than YOLOv5 and significantly surpasses other models. In terms of speed, YOLOv8 can process approximately 45 images per second, nearly three times faster than Faster R-CNN.The innovation of this study lies in the application and optimization of YOLOv8 for herbal medicine identification, demonstrating its advantages in accuracy, speed, and practicality. The research outcomes provide a new technical approach for quality control in herbal medicine, potentially advancing the modernization and standardization of Traditional Chinese Medicine.