Our dataset consists of a pretraining dataset and a fine-tuning dataset. The pretraining dataset is generated using the FDTD method. We simulated a scenario for underground pipeline detection, where the transmitter (tx) is located above the ground, and the receiver (rx) is approximately 30 cm away from the transmitter. The target pipeline buried underground has depths ranging from 1 to 3 meters and a length of approximately 10 meters.