Lightweight Framework for Underground Pipeline Recognition and Spatial Localization Based on Multi-view 2D GPR Images
Published:Dec 25, 2025 05:00
•1 min read
•ArXiv Vision
Analysis
This paper presents a novel framework for detecting underground pipelines using multi-view 2D Ground Penetrating Radar (GPR) images. The core innovation lies in the DCO-YOLO framework, which enhances the YOLOv11 algorithm with DySample, CGLU, and OutlookAttention mechanisms to improve small-scale pipeline edge feature extraction. The 3D-DIoU spatial feature matching algorithm, incorporating geometric constraints and center distance penalty terms, automates the association of multi-view annotations, resolving ambiguities inherent in single-view detection. The experimental results demonstrate significant improvements in accuracy, recall, and mean average precision compared to the baseline model, showcasing the effectiveness of the proposed approach in complex multi-pipeline scenarios. The use of real urban underground pipeline data strengthens the practical relevance of the research.
Key Takeaways
Reference
“The proposed method achieves accuracy, recall, and mean average precision of 96.2%, 93.3%, and 96.7%, respectively, in complex multi-pipeline scenarios.”