Scholarlink

<p>The accurate detection of submarine pipelines and cables is essential for marine energy transmission. Current object detection algorithms exhibit limitations due to poor underwater visibility, complex seabed terrain, and dense occlusions, resulting in high false positive rates and missed detections. This study presents SPC-YOLO (Submarine Pipeline and Cable-YOLO), an enhanced object detection framework that combines YOLOv8 with the ByteTrack tracking algorithm. The framework introduces three key innovations: (1) Enhanced Feature Extraction: Integration of a Diverse Branch Block (DBB) into the backbone network to enhance multi-scale feature representation. (2) Adaptive Feature Learning: Replacement of the original C2f module with an Inverted Residual Mobile Block (iRMB) in the neck section, and implementation of a novel Dual Multi-Scale Attention (DMSA) mechanism for adaptive spatial-contextual feature fusion. (3) Training Optimization: Implementation of a Soft Intersection over Union (SIoU) loss function to improve bounding box regression accuracy. Additionally, the ByteTrack algorithm enables pipeline and cable tracking in video sequences. Extensive experiments on a real-world dataset from side-scan sonar videos in Bohai Bay demonstrate that SPC-YOLO achieves a precision of 93.3%, recall of 89.1%, and mean Average Precision (mAP) of 94.9% and 59.2% at IoU thresholds of 0.50 and 0.50:0.95, respectively. These results represent improvements of 6.9%, 4.1%, 4.6%, and 6.1% compared with YOLOv8, validating SPC-YOLO’s superior detection accuracy and robustness in challenging underwater environments and establishing a framework for surveillance and maintenance of critical seabed infrastructure.</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

SPC-YOLO: A High-Precision Object Detection Method for Submarine Pipeline and Cable

  • 期刊论文

  • 发表日期：

    2026年4月1日

查看全文
  • Qing-jing Cai,
  • Xing-lan Bai

摘要

The accurate detection of submarine pipelines and cables is essential for marine energy transmission. Current object detection algorithms exhibit limitations due to poor underwater visibility, complex seabed terrain, and dense occlusions, resulting in high false positive rates and missed detections. This study presents SPC-YOLO (Submarine Pipeline and Cable-YOLO), an enhanced object detection framework that combines YOLOv8 with the ByteTrack tracking algorithm. The framework introduces three key innovations: (1) Enhanced Feature Extraction: Integration of a Diverse Branch Block (DBB) into the backbone network to enhance multi-scale feature representation. (2) Adaptive Feature Learning: Replacement of the original C2f module with an Inverted Residual Mobile Block (iRMB) in the neck section, and implementation of a novel Dual Multi-Scale Attention (DMSA) mechanism for adaptive spatial-contextual feature fusion. (3) Training Optimization: Implementation of a Soft Intersection over Union (SIoU) loss function to improve bounding box regression accuracy. Additionally, the ByteTrack algorithm enables pipeline and cable tracking in video sequences. Extensive experiments on a real-world dataset from side-scan sonar videos in Bohai Bay demonstrate that SPC-YOLO achieves a precision of 93.3%, recall of 89.1%, and mean Average Precision (mAP) of 94.9% and 59.2% at IoU thresholds of 0.50 and 0.50:0.95, respectively. These results represent improvements of 6.9%, 4.1%, 4.6%, and 6.1% compared with YOLOv8, validating SPC-YOLO’s superior detection accuracy and robustness in challenging underwater environments and establishing a framework for surveillance and maintenance of critical seabed infrastructure.