摘要
在服装定制成衣试穿环节,由于人体体型多样性,经常出现服装试穿不合身的情况。为协助改衣师快速准确找到板型问题及修改方案,以女西裤为例,收集常见的女西裤弊病图像(前裆堆量明显、前裆猫须明显、后片夹裆和后片大腿根堆斜褶明显)作为数据集,并采用深度学习算法中的YOLOv8模型进行实验。研究表明:模型测试阶段,精确度、召回率、平均精度均值(I_(ou)=50%)均达到98%以上,同时结合弊病修正实验验证了弊病修正建议的合理性,实现了女西裤板型弊病的智能检测与修正。
Due to the diversity of human body types,the garments are usually unfit during the trying on process.In order to assist the pattern makers to find and solve the pattern problems,this paper took womens pants as an example,summarized common images of womens trouser malpractices as datasets,including obvious front crotch excess,obvious cat whiskers in the front crotch,pinched crotch in the back piece,and obvious diagonal folds in the thigh-root in the back piece,and conducted experiments using the YOLOv8 model in deep learning algorithms.In the model testing stage,the accuracy,recall ralte,and P_(A)(l_(ou)=50%)all reached over 98%.At the same time,the rationality of the defect correction sugges-tions was verified through defect correction experiments.The study achieved intelligent detection and correction of defects in women's trousers.
作者
彭会齐
陈敏之
PENG Huiqi;CHEN Minzhi(School of Fashion Design and Engineering,Zhejiang Sci-Tech University,Hangzhou 310018,China;School of International Education,Zhejiang Sci-Tech University,Hangzhou 310018,China)
出处
《服装学报》
CAS
北大核心
2024年第1期27-35,共9页
Journal of Clothing Research
关键词
深度学习
目标检测
板型弊病
纸样修正
YOLOv8模型
deep learning
target detection
paper pattern defects
paper pattern correction
YOLOv8 model
作者简介
彭会齐(1992-),女,硕士研究生;通信作者:陈敏之(1978-),女,教授,硕士生导师。主要研究方向为数字化服装。Email:cmz_m@163.com。
