SPHERE

“SPHERE 挑战的数据集包含一个由加速度计、RGB-D 和环境数据组成的多模式活动识别数据集。加速度计以 20 Hz 采样并以其原始格式给出。未给出原始视频是为了保存参与者的匿名性。取而代之的是，提供与被识别人员的质心和边界框相关的提取特征。环境数据由被动红外 (PIR) 传感器组成，这些传感器以原始格式给出。二十（姿势/ambulation) 活动标签在数据集中进行了注释。”

The dataset for the SPHERE Challenge is a multimodal activity recognition dataset composed of accelerometer, RGB-D, and environmental data. The accelerometer data is sampled at 20 Hz and provided in its raw format. Raw video footage is not provided to preserve the anonymity of the participants. Instead, extracted features related to the centroids and bounding boxes of the identified individuals are provided. The environmental data consists of passive infrared (PIR) sensors, which are provided in their raw format. Twenty activity labels corresponding to postural/ambulatory actions are annotated in the dataset.

搜集汇总

背景与挑战

背景概述

SPHERE是一个多模式活动识别数据集，包含加速度计、RGB-D和环境数据，用于20种姿势/移动活动的识别。数据以保护隐私的方式提供，由布里斯托尔大学于2016年发布。