To address the issue of BoShao (Paeonia lactiflora) being frequently substituted with inferior varieties in the traditional Chinese medicine market, this study proposes an enhanced deep learning-based recognition method by refining YOLOv8, aiming to overcome the inefficiency of manual detection and the challenges of small object recognition under complex backgrounds. The proposed approach integrates a dual channel-spatial attention mechanism (CBAM) and a pyramid-like localized patch network (PLPNet) to optimize YOLOv8's feature extraction capabilities: CBAM enhances perception of critical regions through channel-wise weighting and spatial focusing, while PLPNet improves fine-grained feature capture via multi-scale localized patch fusion. Experiments were conducted on a dataset comprising 1,659 field-collected BoShao images, augmented with preprocessing techniques such as random occlusion and Gaussian noise. Ablation experiments demonstrated that the YOLOv8+CBAM+PLPNet model achieved [email protected] and [email protected] scores of 98.66% and 60.70%, respectively, representing improvements of 2.28% and 4.81% over the baseline model. In comparative experiments, the proposed model outperformed state-of-the-art detectors including YOLOv11 and EfficientDet, achieving superior precision (98.69%) and recall (98.18%). The results confirm that the synergistic optimization of attention mechanisms and multi-scale feature fusion significantly enhances robustness in BoShao recognition under complex environmental conditions. This work provides an efficient automated solution for quality inspection of medicinal materials in practical agricultural applications.

Xing Chunyu, Chen Jixiang, Liu Ling, Tan Jun, Yin Wei. Research on BoShao Recognition Based on Deep Learning. Academic Journal of Computing & Information Science(2025), Vol. 8, Issue 3: 28-35. https://doi.org/10.25236/AJCIS.2025.080304.

[1] Ultralytics. (2024). YOLOv8: A new version of YOLO for object detection. GitHub. https://github.com/ultralytics/yolov8

[2] Woo, S., Park, J., Lee, J. Y., & Kweon, I. S. (2018). CBAM: Convolutional Block Attention Module. In Proceedings of the European Conference on Computer Vision (ECCV), 3–19.

[3] Tang, Z., He, X., Zhou, G., Chen, A., Wang, Y., Li, L., & Hu, Y. (2023). A Precise Image-Based Tomato Leaf Disease Detection Approach Using PLPNet. Plant Phenomics, 5, 0042.