基于改进MobileNetV3模型的服装流行色研究
Exploring fashion trends in color through enhanced MobileNetV3 model
刘凤华 1刘兆琪 1刘卫光 1赵红升1
作者信息
- 1. 中原工学院计算机学院 ,河南郑州 450007
- 折叠
摘要
鉴于目前基于权威部门发布数据分析预测服装流行色方法存在的数据集受限、不够精准、数据实时性差等问题,提出了基于改进MobileNetV3模型的服装流行色研究方法.采用改进的M obileNetV3 模型,快速处理服装分类问题;以时序化电商平台销售数据为样本,基于GrabCut算法分析服装图像的主颜色;通过K-means算法统计主颜色和其他颜色的占比;对服装主颜色进行时间维度、服装种类维度和品牌维度的分析,以得出服装流行色的趋势数据.研究发现,相较于传统方法,基于改进MobileNetV3模型的服装流行色研究方法所得数据实时性更强、容量更大,其分析速率也更高.
Abstract
Given the limitations of current methods for analyzing and predicting fashion trends in col-or based on data published by authoritative institutions,such as limited data sets,insufficient accura-cy,and poor real-time data,this paper proposes a research method for fashion color trends based on an improved MobileNetV3 model.The enhanced MobileNetV3 model is employed for rapid clothing classification tasks.Sales data from a time-sequenced e-commerce platform is used as samples,and the main colors of the clothing are analyzed based on the GrabCut algorithm.The proportions of the main colors and mixed colors are calculated using the K-means algorithm.The trend data for fashion trends in color is obtained by analyzing the main colors of clothing from three dimensions:time,clothing type,and brand.The study found that compared to traditional methods,the proposed research meth-od based on the enhanced MobileNetV3 model offers greater real-time data accuracy,increased capaci-ty and quicker analysis speed.
关键词
MobileNetV3/GrabCut/K-means/流行色/主颜色提取Key words
MobileNetV3/GrabCut/K-means/popular color/primary color extraction引用本文复制引用
出版年
2024
NETL
NSTL
万方数据