Scholarlink

We present a novel framework for object detection and segmentation in large-scale 3D point clouds. Our approach integrates edge-aware feature extraction with graph-based clustering to achieve highly accurate semantic segmentation. Unlike conventional methods relying heavily on handcrafted features or extensive labeled datasets for deep learning, our framework leverages geometric consistency and topological constraints to achieve robust object partitioning. We demonstrate the efficacy of our method on benchmark datasets, achieving state-of-the-art precision in complex environments with incomplete data.

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

Advanced Graph-Based Object Segmentation in Large-Scale 3D Point Clouds

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Marcin Mazur,
  • Marcin Daszuta,
  • Dominik Szajerman,
  • Piotr Napieralski

摘要

We present a novel framework for object detection and segmentation in large-scale 3D point clouds. Our approach integrates edge-aware feature extraction with graph-based clustering to achieve highly accurate semantic segmentation. Unlike conventional methods relying heavily on handcrafted features or extensive labeled datasets for deep learning, our framework leverages geometric consistency and topological constraints to achieve robust object partitioning. We demonstrate the efficacy of our method on benchmark datasets, achieving state-of-the-art precision in complex environments with incomplete data.