Labeled inertial sensor data of children with mobility impairments for activity recognition purposes
Description
This dataset comprises inertial sensor data of 31 school-aged children undergoing rehabilitation (see patients_characteristics.mat). These children were able to walk or use a manual wheelchair for household distances. They were equipped with five ZurichMOVE sensor modules. The module comprises a motion processing unit (MPU-9250, InvenSense Inc., San Jose, CA, USA) including a 3-axis accelerometer (raw_data.sensors.acc, g), a 3-axis gyroscope (raw_data.sensors.gyro, °/s), and a 3-axis digital compass (raw_data.sensors.magneto, uT). It further includes an altimeter (MS5611- 01BA01, TE Connectivity Ltd., Schaffhausen, Switzerland; raw_data.sensors.press, mbar) and a Bluetooth® low energy system enabling time synchronization (raw_data.sensors.sync, 0 = synchronization ON | 1 = synchronization OFF). The sensors were placed on both wrists, the sternum, and the thigh and ankle of the less-affected side with corresponding hook-and-loop straps. Additional sensors were placed on assistive devices (e.g., crutches, walking frames, etc.) and the spokes of the wheelchair if applicable. The sensors' position are indicated under raw_data.sensors.header.position. With this equipment, the participants performed a semi-structured activity circuit at the rehabilitation center. They watched a movie on a tablet in their bedroom, played a game of their choice in the living room (e.g., playing cards, doing puzzles, etc.), drank a glass of water in the restaurant, cycled in the gym hall, and played what they wanted to on the outdoor playground (e.g., catching and throwing balls, swinging, etc.). Participants were encouraged to walk, wheel, climb stairs, and take the elevator between these facilities, depending on their functional abilities. No instructions were given on how to perform these activities. Video recordings from an external perspective served as reference criteria. The sampling rates of the camera and the sensors were set to 50 Hz, and timestamps were synchronized with the children clapping their hands in front of the camera. Two researchers labeled the video recordings independently as lying (1.1), sitting (1.2), standing (1.3), kneeling (1.4), being carried (1.5), active wheeling (2.1), passive wheeling (2.2), free walking (3.0), assisted walking (3.1), going upstairs (4.1), going downstairs (4.2), cycling (5), jumping (6), sliding (7), swinging (8), and unknown (not visible, 9). Disagreements lasting more than one second were discussed retrospectively. In the case of consensus, the labels were corrected. Otherwise, the labels were retained. A detailed definition of the individual activities is described in labeling_criteria.pdf.