Efficient Real-Time Bed Occupancy Monitoring Using Toy-Mounted Accelerometer Data

We gather this dataset for toy-basedbed occupancy monitoring. An MPU-6050 IMU is affixed to a plush toy on the bed, and the accelerometer data are collected from the IMU. The accelerometer operates at up to 1 kHz, with a range of ±2 g, a 16-bit ADC, and a noise density of 400 𝜇g/√Hz. Data recorded over two hours, yielding a dataset of 720 10-second samples collected from 5 voluntary subjects for classification among four activities: entering the bed, lying still, turning over, and exiting the bed. The subjects, all of standard body measurements, were instructed to perform these activities without movement pattern restrictions, thereby naturally incorporating movement diversity into the dataset.本数据集专为基于毛绒玩具的床位占用监测任务构建。我们将MPU-6050惯性测量单元（Inertial Measurement Unit, IMU）固定于床上的毛绒玩具上，并采集该惯性测量单元输出的加速度计数据。该加速度计最高采样率可达1 kHz，量程为±2 g，搭载16位模数转换器（Analog-to-Digital Converter, ADC），噪声密度为400 μg/√Hz。本次数据采集总时长为两小时，共从5名自愿受试者处获取720条时长为10秒的样本，用于四类动作的分类任务，四类动作分别为：上床、静卧不动、翻身以及下床。所有受试者均为标准体型，实验要求其在执行上述动作时无动作模式限制，由此自然地将动作多样性融入数据集当中。