Answers are grounded in this paper and cite their sources.
Experimental evaluation demonstrates high-accuracy foreign object classification on power lines using 3D lidar, indicating enhanced automated aerial grid inspection.
This study addresses the classification of foreign objects attached to transmission lines using airborne laser scanning (ALS) point clouds. In practical inspection scenarios, complex backgrounds, sparse point distributions, and limited object categories make robust feature learning difficult. To improve point-cloud representation and classification performance, we propose AP-Net, an attention-enhanced convolutional network with point-cloud preprocessing and 3D data augmentation. The network directly processes ALS point-cloud blocks without converting them into 2D images. It integrates a Point Attention Module (PAM) and a Group Attention Module (GAM) to capture boundary-level discriminative features and object-level structural consistency. Experimental results on 6,000 point-cloud samples, including 3,000 kite samples and 3,000 bird samples, show that AP-Net achieves an average classification accuracy of 96.65%, precision of 96.15%, recall of 96.20%, and F1-score of 96.15%. Ablation experiments further demonstrate that the proposed attention modules improve accuracy by 3.2%, precision by 4.1%, and recall by 3.8% compared with the baseline model without attention modules.
No takes yet. Share an insight, caveat, or question.
Xiao et al. (2026) studied this question.
Synapse has enriched 8 closely related papers on similar clinical questions. Consider them for comparative context: