Scholarlink

<p>In the realm of process mineralogy, the utilization of microscope proves to be more cost-effective than scanning electron microscopy and X-ray diffraction for mineral composition analysis. This paper introduces a novel approach to polished section image segmentation labeling, employing the Mask R-CNN deep learning algorithm. The primary objective is the swift and precise segmentation and labeling of minerals within mineral photometry images. In contrast to the conventional manual identification method, characterized by prolonged time requirements and errors stemming from subjective human judgment, this deep learning methodology not only accomplishes segmentation and annotation within seconds but also attains an impressive accuracy of 89.25%. This establishes a promising groundwork for the future automation of mineral composition analysis.</p>

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

Segmentation and Labeling of Polished Section Images Based on Deep Learning

  • 期刊论文

  • 发表日期：

    2025年2月15日

查看全文
  • Haopo Tang,
  • Lifang He,
  • Bin Huang,
  • Songwei Huang,
  • Guanyu Ma

摘要

In the realm of process mineralogy, the utilization of microscope proves to be more cost-effective than scanning electron microscopy and X-ray diffraction for mineral composition analysis. This paper introduces a novel approach to polished section image segmentation labeling, employing the Mask R-CNN deep learning algorithm. The primary objective is the swift and precise segmentation and labeling of minerals within mineral photometry images. In contrast to the conventional manual identification method, characterized by prolonged time requirements and errors stemming from subjective human judgment, this deep learning methodology not only accomplishes segmentation and annotation within seconds but also attains an impressive accuracy of 89.25%. This establishes a promising groundwork for the future automation of mineral composition analysis.