An Efficient Lightweight Object Detection Algorithm for Defect Identification in Paeoniae Radix Alba With Rapid Area and Diameter Estimation Using OpenCV

As an important traditional Chinese medicinal herb, Paeoniae Radix Alba (white peony root, WP) not only has significant medicinal value in the field of Chinese medicine but is also widely used in food products and health supplements due to its food and medicine dual-use properties. The quality of Paeoniae Radix Alba is crucial for its efficacy, safety, and effectiveness in food applications, making efficient inspection methods essential. The algorithm uses a self-developed detection head (LSD-Head) combined with a Feature Fusion Attention (FCA) mechanism for effective defect detection. Additionally, OpenCV technology is used to accurately measure the physical dimensions of the slices. Compared to the YOLOv8 model (76.4% precision, 63.7% recall, 70.1% mAP@0.5, 8.2 GFLOPs), the proposed model reduces parameter size by 13%, reaching only 65.8% of YOLOv8's size, while improving accuracy by 2.6%. For physical dimensions, the average error between the detected and actual sizes is controlled within 5%. Experimental results show that the proposed algorithm performs well in defect detection and accurately measures the area and maximum and minimum diameters of Paeoniae Radix Alba decoction pieces.