DL-Aided Underground Cavity Morphology Recognition Based on 3D GPR Data

Cavity under urban roads has increasingly become a huge threat to traffic safety. This paper aims study cavity morphology characteristics and proposes deep learning (DL)-based classification method using the 3D ground-penetrating radar (GPR) data. Fine-tuning technology in DL can be used some cases with relatively few samples, but case of only one or very there will still overfitting problems. To address this issue, simple general framework, few-shot (FSL), is first employed for tasks, based on which classifier learns identify new classes given examples. We adopt relation network (RelationNet) as FSL consists an embedding module module. Furthermore, proposed simpler faster because it does not require pre-training fine-tuning. The experimental results are validated GPR road modeling data obtained from gprMax3D system. compared other networks such ProtoNet, R2D2, BaseLine relative different benchmarks. demonstrate that outperforms prior approaches, its average accuracy reaches 97.328% four-way five-shot problem support samples.