Answers are grounded in this paper and cite their sources.
Shows improved antenna classification accuracy in diverse environments, suggesting enhanced asset management efficiency.
As the deployment density of 5G networks grows rapidly, the importance of efficient antenna asset management has become increasingly prominent. However, existing manual inspection methods face clear limitations, including high costs and the risk of safety accidents. This study proposes an integrated system designed to automatically detect antenna objects and classify their detailed attributes from images captured in diverse real-world operational environments. The proposed system adopts a two-stage network architecture where object detection and attribute classification are performed sequentially. It incorporates data augmentation techniques tailored to field-specific characteristics, such as occlusions and varying weather conditions. Furthermore, optimization strategies were introduced to mitigate class imbalance and label ambiguity issues encountered during the training process. Experimental results demonstrate robust performance in actual operating environments, achieving 93.9% in the detection metric (mAP@0.5) and 89.6% in classification (Top-1 Accuracy). The findings of this research suggest the practical feasibility of streamlining asset management processes through image-based automated antenna identification.
No takes yet. Share an insight, caveat, or question.
Choi et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: