基于YOLOv5的服装熨烫目标检测算法研究  

Research on Clothing Ironing Target Detection Algorithm Based on YOLOv5

作　　者：姚明杰 李斌 陈世海 李晓帆 麻方达 符朝兴[1] YAO Mingjie;LI Bin;CHEN Shihai;LI Xiaofan;MA Fangda;FU Chaoxing(College of Mechanical and Electrical Engineering,Qingdao University,Qingdao 266071,China)

机构地区：[1]青岛大学机电工程学院,山东青岛266071

出　　处：《青岛大学学报（工程技术版）》2023年第1期24-33,共10页Journal of Qingdao University(Engineering & Technology Edition)

摘　　要：针对服装熨烫行业中熨烫目标分类模糊、人工方式导致检测不准确且效率低的问题,本文将YOLOv5算法运用到服装行业熨烫目标检测中,将常见的熨烫目标分为裤兜、缝线及褶皱,建立对应的数据集并标注。同时,通过数据集训练YOLOv5算法模型,对模型进行评价和测试,得到模型准确率达98%,召回率达97%,平均精度均值达95%。同时,选择200张熨烫目标图像,对模型进行测试实验。实验结果表明,该模型对裤兜、缝线和褶皱的识别率分别为100%,96%和95%,检测置信度为0.82~0.97,检测效果较好,满足实验要求,实现了常见服装熨烫目标的识别、分类及定位。该研究提升了服装熨烫行业的生产效率。Aiming at the problem that the ironing targets in the clothing ironing industry are fuzzy in classification and detection is mainly caused by manual methods,which leads to inaccurate detection and low efficiency,this paper applies YOLOv5 algorithm to the detection of ironing targets in the clothing industry.Common ironing targets are divided into trouser pocket,seam and fold,and corresponding data sets are established and labeled.At the same time,the YOLOv5 algorithm model is trained through the dataset,and the model is evaluated and tested.The accuracy of the model reaches 98%,the recall rate reaches 97%,and the average accuracy reaches 95%.Meanwhile 200 ironing target images to test the model.The experimental results show that the recognition rate of trouser pockets,stitches are selected and folds is 100%,96%and 95%,respectively,and the detection confidence is 0.82~0.97.The detection effect of the model is good,and the detection effect of the ironing target meets the experimental requirements,and the recognition,classification and location of common clothing ironing target are realized.The research has boosted productivity in the garment ironing industry.

关 键 词：YOLOv5 目标检测 服装熨烫 深度学习 

分 类 号：TP391.412[自动化与计算机技术—计算机应用技术]