Scholarlink

<p>Ensuring the safe and efficient operation of pipelines is crucial for the natural gas transportation system. Pipeline leaks pose significant risks and must be detected and addressed promptly. To overcome the limitations of existing algorithms, such as susceptibility to background interference, challenges in effectively fusing multi-scale feature information, and difficulty detecting small leaks, we propose the RCT-YOLO model. This model incorporates the C3K2_RVBE module to enhance feature adaptation and fusion. In order to tackle the challenges of multiscale feature fusion in gas leak detection, we introduce the CRSGFPN module, which significantly improves feature fusion capabilities. Additionally, the lightweight detector head reduces the number of parameters while enhancing small target detection performance. Experimental results demonstrate that RCT-YOLO achieves 60% mAP@0.5 and outperforms the original YOLOv11. Specifically, precision, recall, mAP@0.5, and mAP@0.5–0.95 are improved by 5.1%, 1.7%, 3.9%, and 12.3%, respectively. Furthermore, the model performs excellently on the Kaggle pipeline leakage detection dataset and VisionDrone2019. A comparative analysis shows that RCT-YOLO surpasses other mainstream models in both detection accuracy and computational efficiency, highlighting its strong potential for natural gas pipeline leakage monitoring applications.</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

Natural gas pipeline leak detection algorithm based on semantic reconstruction of contextual rectangular regions

  • 期刊论文

  • 发表日期：

    2025年9月1日

查看全文
  • Jiaxin Luo,
  • Zifeng Liu,
  • Sergey Aleksandrovich Martynov,
  • Xuliang Zhao,
  • Jianping Tan,
  • Siyu Lu

摘要

Ensuring the safe and efficient operation of pipelines is crucial for the natural gas transportation system. Pipeline leaks pose significant risks and must be detected and addressed promptly. To overcome the limitations of existing algorithms, such as susceptibility to background interference, challenges in effectively fusing multi-scale feature information, and difficulty detecting small leaks, we propose the RCT-YOLO model. This model incorporates the C3K2_RVBE module to enhance feature adaptation and fusion. In order to tackle the challenges of multiscale feature fusion in gas leak detection, we introduce the CRSGFPN module, which significantly improves feature fusion capabilities. Additionally, the lightweight detector head reduces the number of parameters while enhancing small target detection performance. Experimental results demonstrate that RCT-YOLO achieves 60% mAP@0.5 and outperforms the original YOLOv11. Specifically, precision, recall, mAP@0.5, and mAP@0.5–0.95 are improved by 5.1%, 1.7%, 3.9%, and 12.3%, respectively. Furthermore, the model performs excellently on the Kaggle pipeline leakage detection dataset and VisionDrone2019. A comparative analysis shows that RCT-YOLO surpasses other mainstream models in both detection accuracy and computational efficiency, highlighting its strong potential for natural gas pipeline leakage monitoring applications.