A Raw UWB Radar Dataset for Human Detection

In this dataset, a human detecting model using with UWB radar technology is presented. Two distinct datasets were created using the UWB radar device, leveraging its dual features. Data collection involved two main scenarios, each containing multiple sub-scenarios. These sub-scenarios varied parameters like the position, distance, angle, and orientation of the human subject relative to the radar. Unlike conventional approaches that rely on signal processing or noise/background removal, this study uniquely emphasizes analyzing raw UWB radar data directly. 

In the dataset, two human subject that is male and female have been participated for data collection. The male participant is 25 years old and 170cm tall, and the female participant is 17 years old and 167cm tall. There were no obstacles between the radar and the human subjects. Each sample was recorded for one minute.  The experiment setup was set up in an indoor environment and two different dataset have been created. The datasets named as Dataset-1 and Dataset-2. In the Dataset-1, when the subjects position is standing, the UWB radar was on a table 1.5 meters and when subjects position is sitting, the UWB radar was on a table 1.2 meters  from the ground at the same level as a human chest. In the Dataset-2, the subjects position was only standing and the UWB radar was on a table 1.5 meters from the ground at the same level as a human chest.

For the collection of Dataset-1, the subjects were in sitting and standing positions. The data collection consisted of different scenarios involving the human subjects, where they standing and sitting at various angles (30°, 45° and 90°) and distances (10 cm to 100 cm, in 10 cm intervals) from the radar. Each position was measured three times, resulting in a detailed capture of radar reflections at different spatial configurations. When the subjects are in front of radar at 90° the distance is from 10 cm to 100cm in standing and sitting positions. Otherwise when the subjects are right and left side of radar at 30° and 45°, the distances are 20 cm, 50 cm and 100 cm in standing and sitting positions. Additionally, a non-human scenario was recorded, where the room was with chair and without chair. In this setup, when the room is with chair, the angles, distances, directions is same as involving human subjects scenarios. When the room is without chair, the distances, angles and directions has not been considered. Totally 685 samples were collected in the seated measurements, with 500 containing a human and 185 without a human. 

Dataset-2 was collected in the same indoor environment and with the same two human subjects. The experimental setup and radar positioning mirrored those in Dataset-1, with the key difference not being the inclusion of recordings with the human subjects in a sitting position.  This dataset also contained samples with the human subject standing, which followed the same distance and angle variations as in Dataset-1 but the interval of distances for each samples were ten. Additionally, a non-human scenario, the angles, directions, distances and interval are same as human subjects scenarios. In total, 1540 samples were recorded, with 1040 samples containing a human subject and 500 samples without a human.