基于深度学习的汽车前脸家族化造型设计研究 ——以奥迪品牌汽车为例
摘 要：在设计学领域，专家经验通常是分析与提取汽车品牌家族化造型特征的主要依据，为了减少人工介入，实现智能化分析与提取家族化造型特征，文章以奥迪汽车品牌为例进行探索。研究通过构建奥迪汽车前脸图像数据库，采用深度学习卷积神经网络作为特征识别的核心算法，利用类别激活映射对结果进行可视化，同时开发高效、智能的品牌特征提取分析模型，实现造型基因的智能分析与提取。该方法从艺科融合的角度改变了传统依赖主观经验和手动分析的局限，以期提高设计工作的效率，提升当前汽车品牌家族化造型特征分析的技术水平。
关键词：工业设计；家族化造型设计；深度学习；汽车造型；奥迪品牌
中图分类号：U462 文献标识码：A 文章编号：1672-7053(2024)05-0029-04
基金项目：教育部人文社会科学研究规划基金项目“基于深度学习的汽车品牌家族化造型基因智能分析方法研究”（22YJA760072）
Abstract：In the field of design, expert experience is usually the main basis for analyzing and extracting family-oriented styling features of automobile brands. In order to reduce manual intervention and realize intelligent analysis and extraction of family-oriented styling features, this paper takes Audi automobile brand as an example to explore. By constructing the front face image database of Audi cars, deep learning convolutional neural network is adopted as the core algorithm of feature recognition, and category activation mapping is used to visualize the results. An efficient and intelligent brand feature extraction analysis model is developed to realize intelligent analysis and extraction of modeling genes. This method changes the traditional limitation of relying on subjective experience and manual analysis from the perspective of art integration, in order to improve the efficiency of design work and improve the technical level of current automobile brand family modeling feature analysis.
Key Words：Industrial Design; Family-oriented Styling Design; Deep Learning; Automobile Styling; Audi Brand
