基于改进YOLO11的现代汉服裙装形制识别算法OA

To address the high cost and low efficiency of manual annotation for garment styles in the Hanfu e-commerce industry,a modern Hanfu skirt style recognition algorithm based on an improved YOLO11 was pro-posed.Firstly,five major categories of Hanfu skirt styles were summarized and a corresponding data set was con-structed.Secondly,to address the limitations of the YOLO11 model in detecting Hanfu skirt styles,the backbone network was enhanced by incorporating DualConv and Focal Modulation modules.The SEAM module was embed-ded into the neck network and the ATFL was introduced to optimize the classification loss function.Finally,abla-tion experiment and model comparison experiments were conducted to evaluate the improvements.Experimental results demonstrated that compared with YOLO11n,the improved model achieved increases of 2.63%in preci-sion,2.45%in recall,0.94%in mean Average Precision(mAP)at an Intersection over Union(IoU)threshold of 0.5,and 1.21%in mAP at IoU thresholds from 0.5 to 0.95,achieving a balance between model lightweight-ing and detection accuracy and offering a reference for Hanfu style classification and recognition.