Scholarlink

Understanding floor image data is important for planning, 3D construction. In this paper, we propose a novel approach for floor content recognition and implementation with Img2Vec, which is converts images into vectors. The proposed method can obtain better performance with 3 key steps: character erasion, corner detection, corner re-order. The implemented system can process input floor image into vector file and corresponding coordinate to generate a certain polygon.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

An Img2Vec Based Floor Recognition and Implementation Method

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Zhiqing Zhang,
  • Jing Li,
  • Liang Yan,
  • Shulin Wu,
  • Xu Zhang

摘要

Understanding floor image data is important for planning, 3D construction. In this paper, we propose a novel approach for floor content recognition and implementation with Img2Vec, which is converts images into vectors. The proposed method can obtain better performance with 3 key steps: character erasion, corner detection, corner re-order. The implemented system can process input floor image into vector file and corresponding coordinate to generate a certain polygon.