Lightweight Framework for Underground Pipeline Recognition and Spatial Localization Based on Multiview 2-D GPR Images

To address the issues of weak correlation between multiview features, low recognition accuracy of small-scale targets, and insufficient robustness in complex scenarios in underground pipeline detection using 3-D ground penetrating radar (GPR) (3D GPR), this article proposes a 3-D pipeline intelligent detection framework that integrates multistrategy improved deep learning (DL) technology. This article explores a novel pathway to achieve accurate 3-D localization through lightweight joint analysis of multiview 2-D GPR images. First, based on a B/C/D-scan three-view joint analysis strategy, a 3-D pipeline three-view feature evaluation method is established by cross-validating forward simulation results obtained using finite-difference time-domain (FDTD) methods with actual measurement data. Second, the dysample CGLU outlookattention-you only look once (DCO-YOLO) framework is proposed, which integrates DySample dynamic upsampling, convolutional gate linear unit (CGLU), and OutlookAttention cross-dimensional correlation mechanisms into the original YOLOv11 algorithm, significantly improving the small-scale pipeline edge feature extraction capability. Furthermore, a 3D-distance intersection over union (DIoU) spatial feature matching algorithm is proposed, which integrates 3-D geometric constraints and center distance penalty terms to achieve automated association of multiview annotations. The three-view fusion strategy resolves inherent ambiguities in single-view detection. Experiments based on 100 km of real urban underground pipeline data show that the proposed method achieves accuracy, recall, and mean average precision of 96.2%, 93.3%, and 96.7%, respectively, in complex multipipeline scenarios, which are 2.0%, 2.1%, and 0.9% higher than the baseline model. Ablation experiments validated the synergistic optimization effect of the dynamic feature enhancement module, and Grad-CAM++ heatmap visualization demonstrated that the improved model significantly enhanced its ability to focus on pipeline geometric features. This study integrates DL optimization strategies with the physical characteristics of 3-D GPR, offering an efficient and reliable novel technical framework for the intelligent recognition and localization of underground pipelines.