Composable activities dataset

可组合活动数据集由 693 个视频组成，其中包含由 14 个演员执行的 16 个类别的活动。每个活动由 3 到 11 个原子动作组成。使用 Microsoft Kinect 传感器捕获每个序列的 RGB-D 数据，并估计相关身体关节的位置。该数据集提供每个视频的活动注释以及每个视频中每个帧的四个人体部分（左/右臂和腿）中的每一个的动作。The Composable Activities Dataset consists of 693 videos, which encompass activities from 16 categories performed by 14 actors. Each activity is composed of 3 to 11 atomic actions. RGB-D data of each sequence was captured using a Microsoft Kinect sensor, and the positions of relevant body joints were estimated. This dataset provides activity annotations for each video, as well as action annotations for each of the four human body parts (left/right arms and legs) in every frame of the video.

搜集汇总

背景与挑战

背景概述

Composable activities dataset包含693个视频，涵盖16个活动类别，由14位演员执行，每个活动由3至11个原子动作组成。数据通过Microsoft Kinect传感器捕获，提供RGB-D信息、关节位置估计，以及每帧中人体部分的活动注释。该数据集由智利天主教大学·北方大学于2014年发布。