SLP

我们为实验设置了两个评估环境，一个在实验室环境中变成了常规卧室，另一个在东北大学健康科学系的模拟病房中。使用IRB批准的协议，我们从109名参与者 (102名来自家庭环境，7名来自医院) 收集了姿势数据，要求他们躺在床上，并在仰卧、左侧和右侧三个主要类别下随机改变姿势。对于每个类别，收集15个姿势。以4种模式中的每一种方式收集用于家庭设置的总体13,770姿势样本和用于医院设置的945样本。此外，我们将覆盖条件从揭开，覆盖一个 (约1毫米厚度的薄板)，然后覆盖两个 (约3毫米厚度的厚毯子)。在每种掩护条件下，我们在现实生活中的监视场景中等待大约10-20秒来模仿稳定的姿势。我们遵循14个关节姿势的定义。

We established two evaluation environments for the experiment: one was a standard bedroom configured in a laboratory setting, and the other was a simulated ward located in the Department of Health Sciences of Northeastern University. Using an IRB-approved protocol, we collected pose data from 109 participants (102 from the home environment and 7 from the hospital setting). Participants were instructed to lie in bed and randomly change their postures under three main posture categories: supine, left lateral, and right lateral positions. For each of these three categories, 15 postures were collected per participant. Data collection was conducted under each of four modalities, yielding a total of 13,770 posture samples for the home environment and 945 samples for the hospital setting. Additionally, three covering conditions were tested: no covering, one layer (a thin sheet with an approximate thickness of 1 mm), and two layers (a thick blanket with an approximate thickness of 3 mm). Under each covering condition, we waited for approximately 10–20 seconds in a real-world surveillance scenario to simulate stable postures. We adhered to the standardized definition of 14-joint postures.

搜集汇总

背景与挑战

背景概述

SLP数据集包含来自109名参与者的姿势数据，收集于实验室和模拟病房环境，涵盖仰卧、左侧和右侧三个主要姿势类别，共14,715个样本。数据集特别关注了不同覆盖条件下的姿势变化，适用于床上姿势估计研究。