BUT Gait raw dataset

"BUT Gait raw dataset" is a hundred-person dataset on human gait, containing recordings from two motion tracking sessions conducted over two days. The creation of the dataset was necessitated by a specific domain gap that presented a challenge in the validation of biometric methods of human gait collected using inertial sensors for cross-day configurations. The acquisition process utilised the Perception Neuron v32 inertial motion tracking system, which possesses the capability to measure accelerometer and gyroscope values, in addition to estimating information regarding limb and sensor orientation at a frequency of 120 Hz. Furthermore, these values are synchronised in time. The acquisition process necessitated the donning of a dedicated suit equipped with 17 motion sensors, in conjunction with the execution of a calibration procedure. This is one of the few datasets that contains information collected over two days (enabling cross-day validation) and was created for biometric experiments. In the experiment, participants were instructed to walk naturally on a solid, hard surface for a distance of 3 metres. The type of footwear worn remained consistent between data acquisition sessions. The following discussion will explore the factors that ensure the laboratory nature of the tests. During the acquisition process, measurement data was collected and synchronised with 17 motion sensors. The available locations for sensors include the right and left wrists. In the context of proprietary experiments, particular attention was directed towards the right thigh area, whose data exhibited a high degree of similarity to the measurements recorded by MEMS sensors built into mobile phones and placed in trouser pockets.