Scholarlink

<p>An efficient clothing recommendation system can provide users with personalized and intelligent matching suggestions to meet their different needs and preferences. Currently, mainstream algorithms usually recommend best-selling apparel based on the user’s browsing history, purchase records and other information, however, the matching relationship between the clothing itself and the user portrait is not fully considered. Aiming at the above problems, this paper proposes a personalized clothing matching recommendation algorithm based on user portrait. First, the MobileNetV3 network integrated with migration learning is used to classify the user’s facial color, and the AB-CNN network is designed to classify the user’s age and face shape to establish a user objective attribute recognition model. Second, the face recognition technology is used to recognize the user and obtain the user’s history information, and the subjective preference model is designed to construct a high-precision user portrait by combining collaborative filtering and contextual information. Third, the theory of clothing matching is used to calculate the comprehensive score of clothing matching, and accurate recommendations are obtained through user preference feedback to improve user satisfaction. Finally, experiments are carried out on datasets such as AGFW-v2 [50], Polyvore [44], DressCode [41], and the experimental results showed that the maximum values of first match recommendation accuracy, system comprehensive performance, recommendation accuracy rate, recall rate and F1 value reached 88.47%, 96.35, 72.45%, 69.92% and 71.07%, respectively, which was superior to the existing algorithm.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Personalized clothing matching recommendation algorithm based on user portrait

  • 期刊论文

  • 发表日期：

    2025年7月3日

查看全文
  • Xueping Su,
  • Yihang Lei,
  • Can Wang,
  • Yunhong Li,
  • Yingxuan Ren

摘要

An efficient clothing recommendation system can provide users with personalized and intelligent matching suggestions to meet their different needs and preferences. Currently, mainstream algorithms usually recommend best-selling apparel based on the user’s browsing history, purchase records and other information, however, the matching relationship between the clothing itself and the user portrait is not fully considered. Aiming at the above problems, this paper proposes a personalized clothing matching recommendation algorithm based on user portrait. First, the MobileNetV3 network integrated with migration learning is used to classify the user’s facial color, and the AB-CNN network is designed to classify the user’s age and face shape to establish a user objective attribute recognition model. Second, the face recognition technology is used to recognize the user and obtain the user’s history information, and the subjective preference model is designed to construct a high-precision user portrait by combining collaborative filtering and contextual information. Third, the theory of clothing matching is used to calculate the comprehensive score of clothing matching, and accurate recommendations are obtained through user preference feedback to improve user satisfaction. Finally, experiments are carried out on datasets such as AGFW-v2 [50], Polyvore [44], DressCode [41], and the experimental results showed that the maximum values of first match recommendation accuracy, system comprehensive performance, recommendation accuracy rate, recall rate and F1 value reached 88.47%, 96.35, 72.45%, 69.92% and 71.07%, respectively, which was superior to the existing algorithm.