主办：陕西省汽车工程学会
ISSN 1671-7988  CN 61-1394/TH
创刊：1976年
汽车实用技术

汽车实用技术 ›› 2025, Vol. 50 ›› Issue (19): 62-67.DOI: 10.16638/j.cnki.1671-7988.2025.019.011

• 设计研究 • 上一篇    

基于特征识别的汽车仪表板建模方法研究

童少尉 1，马祥磊 1，王辉 2*，邱星辉 2   

  1. 1.广州汽车集团股份有限公司 汽车工程研究院； 2.北京经纬恒润科技股份有限公司
  • 发布日期:2025-10-09
  • 通讯作者: 王辉
  • 作者简介:童少尉（1994－），男，硕士，工程师，研究方向为车内饰结构设计与仿真分析； 通讯作者：王辉（1994－），男，硕士，研究方向为基于图像的特征识别算法研究

Research on Modeling Method for Automotive Dashboard Based on Feature Recognition

TONG Shaowei1 , MA Xianglei1 , WANG Hui2* , QIU Xinghui2   

  1. 1.Automotive Engineering Institute, Guangzhou Automobile Group Company Limited; 2.Beijing Jingwei Hirain Technology Company Limited
  • Published:2025-10-09
  • Contact: WANG Hui

PDF

摘要： 随着汽车行业的快速发展，仪表板设计面临效率与创新的双重挑战。文章提出一种基 于 YOLOv8 算法的汽车仪表板特征识别与快速建模方法，旨在结合历史设计经验的基础上在 仪表板设计初期进行快速原型设计，缩短整体设计周期，提升设计效率。首先，针对现有设 计流程的复杂性，通过造型面与断面生成技术构建仪表板基面；其次，利用 YOLOv8 算法高 效识别仪表板骨架的关键特征，在保证高准确率的同时显著提升识别速度；最后，基于识别 结果调用匹配的特征模块，并将其与基面模型集成，快速生成符合设计需求的仪表板原型。 实验结果表明，训练的特征识别模型准确率和召回率可达到 0.75 和 0.74，mAP@0.5 和 mAP@0.5:0.95 稳定在 0.82 和 0.6 附近。基于识别模型进行仪表板的快速原型设计不仅能提升 设计效率，同时确保了设计的准确性与创新性，为汽车仪表板设计提供了新的技术路径。

关键词: YOLOv8；汽车仪表板；特征识别；快速建模；原型设计

Abstract: With the rapid advancement of the automotive industry, dashboard design confronts dual challenges in efficiency and innovation. This paper proposes a method of dashboard feature recognition and rapid modeling based on YOLOv8 algorithm, aiming to integrate historical design expertise to enable rapid prototyping during the initial design phase, thereby shortening the development cycle and improving design efficiency. First, dashboard base surfaces are constructed using styling surface and cross-section generation techniques to address the complexity of current design workflows. Second, the YOLOv8 algorithm is leveraged to efficiently identify critical features of dashboard skeletons, ensuring high accuracy while substantially accelerating recognition speed. Finally, matched feature modules are dynamically invoked based on recognition results and integrated with base surface models to swiftly generate design-compliant dashboard prototypes. Experimental results indicate that the trained recognition model achieves precision and recall rates of 75% and 75%, respectively, with mAP@0.5 and mAP@0.5:0.95 stabilized at 0.82 and 0.6. The rapid prototyping of dashboard based on the identification model can not only improve the design efficiency, but also ensure the accuracy and innovation of the design, which provides a new technical path for the design of automotive dashboard.

Key words: YOLOv8; automotive dashboard; feature recognition; rapid modeling; prototype design