基于YOLOv8的女裤结构特征识别方法

  • 打印
  • 收藏
收藏成功


打开文本图片集

关键词:女裤结构特征;YOLOv8;女裤制版;款式图识别;相似款式图匹配中图分类号:TS941.71 文献标志码:A 文章编号:1673-3851(2025)07-0507-08引用格式:,.基于YOLOv8的女裤结构特征识别方法[J].学报(自然科学),2025,53(4) :507-514.Reference Format: LI Meng, CHEN Minzhi.A structural feature recognition method of women's pantsbased onYOLOv8[J]. Journal of Zhejiang Sci-Tech University,2025,53(4) :507-514.

Abstract: To facilitate the matching of clothing style sketches and the swift generation of template maps,so as to improve the pattern-making efficiency of apparel enterprises,a YOLOv8-based method for identifying structural features of women's pants was proposed. This method matches similar style sketches by recognizing the structural features of women's pants. According to patern-making rules for women's pants,a clasification system for structural features was established, encompassing six key modules: silhouette,waist structure, hip structure,crotch structure, placket,and pockets,which were further subdivided into 23 categories.These structural features were then encoded according to the classification system to obtain category labels. Taking mannequin-based style sketches as the research object, samples of women's pants were drawn. By formulating and unifying sample annotation rules,a dataset suitable for structural feature recognition was created. The enhanced dataset of style sketches was recognized using YOLOv8,and the model was applied to match similar style sketches and rapidly generate pattern drawings. The results indicate that this method can effectively classify the structural features of women's pants,with an overall classification accuracy of 97.8% and a mean average precision of 99.0% . It provides an effective approach for the recognition and application of garment structural features.

Key words: structural characteristics of women' s pants; YOLOv8;women's pants plate making pattern recognition; similar style sketch matching

0 引言

传统的服装制版需要制版师人工分析款式图信息并绘制样板图,容易受主观判断的影响,造成款式特征误判、款式图到样板图的转换差异大以及生产效率低等问题[1]。(剩余6468字)

monitor