基于改进MobileNetV3模型的服装流行色研究

Exploring fashion trends in color through enhanced MobileNetV3 model

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MobileNetV3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。

Given the limitations of current methods for analyzing and predicting fashion trends in color based on data published by authoritative institutions,such as limited data sets,insufficient accuracy,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 method based on the enhanced MobileNetV3 model offers greater real-time data accuracy,increased capacity and quicker analysis speed.