数码设计Issue(15)：26-28,3.
基于视觉理解的前端页面自动生成技术研究
Research on Visual Understanding-Based Frontend Page Auto-Generation Technology
摘要
Abstract
Addressing the high time cost in frontend page auto-generation practices,this study proposes a visual understanding-based frontend page auto-generation technology.By integrating computer vision and deep learning,the technology constructs a comprehensive technical framework comprising a visual interaction terminal,intelligent configuration center,data service middleware,dynamic template rendering engine,and visual understanding-assisted generation unit.Vision Transformer(ViT)is employed for visual feature encoding of GUI prototypes,combined with a bidirectional LSTM network for semantic decoding.Syntactically compliant front-end code is ultimately generated through autoregressive methods.Experimental results demonstrate that on a dataset of 15,328 samples expanded to a 42-token vocabulary,the average generation time is only 105.01ms.This research provides a novel technical pathway for front-end development automation.关键词
视觉理解/前端页面/自动生成/GUI原型图/深度学习Key words
visual understanding/front-end pages/automated generation/GUI prototypes/deep learning分类
信息技术与安全科学引用本文复制引用
陈勇杰..基于视觉理解的前端页面自动生成技术研究[J].数码设计,2025,(15):26-28,3.基金项目
武汉生物工程学院2024年第一批校级科研项目,融合注意力监测的在线教学分析系统设计与实现(项目编号:24ZKY147). （项目编号:24ZKY147）
