Scholarlink

Surface crack detection is crucial for ensuring the structural integrity of various materials, including metals, concrete, plastics, and composites. Cracks are often the root cause of structural failure, as they allow moisture and other reactive elements to penetrate, which accelerates crack propagation. To mitigate these risks, early detection of surface cracks is essential. This paper proposes an ensemble classification method for surface crack detection, utilizing three Convolutional Neural Networks (CNNs). These individual models are combined through hard voting, a technique that aggregates their outputs to improve overall accuracy. The results demonstrate that the developed model is highly effective in detecting surface cracks, achieving an accuracy of 99.85%, precision of 0.9977, recall of 1.00, and an F1 score of 0.9985. This approach offers significant potential for enhancing the reliability of structural health monitoring systems.

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

Ensemble Deep Learning CNN Models for Accurate Surface Crack Detection

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Jatin Sadarang,
  • Dheeraj Kumar Ghaghre,
  • Nitin Kumar Shrivastava

摘要

Surface crack detection is crucial for ensuring the structural integrity of various materials, including metals, concrete, plastics, and composites. Cracks are often the root cause of structural failure, as they allow moisture and other reactive elements to penetrate, which accelerates crack propagation. To mitigate these risks, early detection of surface cracks is essential. This paper proposes an ensemble classification method for surface crack detection, utilizing three Convolutional Neural Networks (CNNs). These individual models are combined through hard voting, a technique that aggregates their outputs to improve overall accuracy. The results demonstrate that the developed model is highly effective in detecting surface cracks, achieving an accuracy of 99.85%, precision of 0.9977, recall of 1.00, and an F1 score of 0.9985. This approach offers significant potential for enhancing the reliability of structural health monitoring systems.