Automatic Classification System of Chinese Medicinal Materials Based on Visual Perception and Feature Extraction

DOI: https://doi.org/10.62517/jbdc.202501403

Author(s)

Jia Weng1, Wanying Zhang2, Chun Yang2,*

Affiliation(s)

1School of Artificial Intelligent, Guangzhou Huashang College, Guangzhou, Guangdong, China 2School of Accounting, Guangzhou Huashang College, Guangzhou, Guangdong, China *Corresponding Author

Abstract

Chinese medicinal herbs are diverse and complex in form, with some varieties being similar, which can easily lead to confusion and misidentification, affecting the quality of herbs and clinical safety. Traditional identification methods rely heavily on expert experience, making them subjective, inefficient, and unable to meet modern demands. This paper proposes an automatic classification system for Chinese medicinal herbs based on convolutional neural networks, integrating modules for image acquisition, preprocessing, feature extraction, and classification recognition. By constructing a variety of Chinese medicinal herb image datasets and employing optimization strategies such as transfer learning, feature fusion, and data augmentation, the accuracy, generalization ability, and robustness of the system have been significantly improved. Experimental results show that the system achieves an accuracy rate of 95.5% in classification tasks, demonstrating broad application potential for areas such as herb quality inspection, market regulation, and education and training.

Keywords

Chinese Herbal Medicine Identification; Image Classification; Deep Learning; Convolutional Neural Network; Visual Perception

References

[1]Lv W, Kou F, Wang G, et al. Research on road surface perception strategy for intelligent suspension based on vision and dynamics fusion. Measurement Science and Technology, 2025, 36 (9): 096118-096118. [2]Ashurov A, Qu H. An efficient industrial defect detection based on hybrid residual attention with modified generative adversarial network and convolutional neural network model. Computers and Electrical Engineering, 2025, 127 (PA): 110580-110580. [3]Wanqing G, Ning C, Guojiang X, et al. TCM2Vec: a detached feature extraction deep learning approach of traditional Chinese medicine for formula efficacy prediction. Multimedia Tools and Applications, 2023, 82 (17): 26987-27004. [4]Chen B, Yu J, Feng H, et al. Adaptive Cross-Modal Non-Invasive Detection of Crucian Carp Freshness Based on Flexible Haptics and Multi-Angle Visual Perception. Food and Bioprocess Technology, 2025: 1-16. [5]Zhu X, Pang G, He X, et al. A segmentation-combination data augmentation strategy and dual attention mechanism for accurate Chinese herbal medicine microscopic identification. Frontiers in Plant Science, 2024, 151442968-1442968. [6]Hou Q, Yang W, Liu G. Chinese herbal medicine recognition network based on knowledge distillation and cross-attention. Scientific Reports, 2025, 15 (1): 1687-1687. [7]Hu K, Lin H C, Jin X, et al. Meta-YOLOv8: multi-scale few-shot object detection for Chinese medicinal decoction pieces. Journal of King Saud University Computer and Information Sciences, 2025, 37 (7): 201-201. [8]Wang T, Chen J, Liu C, et al. Identification of H2O ice at lunar south polar region based on deep learning. Icarus, 2025, 441116682-116682. [9]Shiney S T, Jerome A S. Optimized attention Induced multi head convolutional neural network with Densenet201 for cervical cancer diagnosis. Biomedical Signal Processing and Control, 2025, 110(PB):108166-108166. [10]Chen Z, Lin Z, Wan H, et al. Rapid and accurate metabolite identification of traditional Chinese medicine based on UPLC-Q-TOF-MS coupled with UNIFI analysis platform and quantitative structure-retention relationship: Danshen-Honghua herbal pair as an example. Journal of Pharmaceutical and Biomedical Analysis, 2025, 257116696-116696.