Acoustic and Seismic Dataset for Mobile Target Recognition

In this paper, a multi-terrain noise environment dataset (ASD) is constructed for noisy conditions (distant highway noise and train trip noise). The dataset contains three types of noise: environmental noise (natural sound and sensor vibration), human activity noise (human voice\/footsteps) and wheeled vehicle noise (engine\/driving noise), and was collected by deploying sensors through both soft ground pinning and hard ground surface placement. The total data amounted to 17,395 seconds (training set: 5,639 seconds for environment, 6,585 seconds for human activities, 5,171 seconds for vehicles; test set: 357 seconds for environment, 343 seconds for human activities, 358 seconds for vehicles), normalized to a unit of 1 second. unit normalized segmentation.