瀚岳人体行为数据集 HanYue-3D Actions
瀚岳人体行为数据集通过Kinect v2 收集了15种简单动作,包括:打电话,喝水,挥手,看手表,拍打衣服上的灰,摔倒,推椅子,原地跳,站起来,坐着不动,站立并鼓掌,走,坐下去,坐着不动,坐着鼓掌;以及4种复杂行为动作,包括:“坐着不动→站起来→拍打身上的灰”,“原地跳→站着不动→原地跳→挥手”,“走→站着不动→原地跳→挥手→坐下”和“坐着鼓掌→站起来→站着鼓掌→原地跳→挥手”。每组动作有年龄在17~21岁之间,身高1.59~1.85米之间的7名男生和2名女生重复3~4次。数据集一共包括简单动作样本:413个,简单动作每组样本数在35~37之间;复杂行为样本共36个,每组9个,复杂样本不参与训练,只用于动作的时序定位实验。
HanYue-3D is a self-collected dataset. It was collected using a Kinect v2.0 camera. It includes 15 simple action types: make a phone call, drink, wave hands, look at a watch, pat dust off clothes, fall, push a chair, jump in place, stand up, stand still, stand clap, walk, sit, sit still, and sit clap. Nine subjects were asked to perform these 15 activities 3 or 4 times. The positions in 3D coordinates of all 25 joints provided by the Kinect v2.0 sensor were recorded. There are 413 samples in total, and each action type is represented by 35–37 samples. Additionally, for TAL purposes, 4 complex behaviour types were collected. Each complex behaviour consists of several of the above simple actions. The 4 complex behaviours are “sit still→stand up→pat dust off clothes”, “jump→stand still→jump→wave hands”, “walk→stand still→jump→wave hands→sit down” and “sit clap→stand up→stand clap→jump→wave hands”.
点击此处下载瀚岳人体行为数据集
click here to download HanYue-Action3D
其它数据集(含HanYue-3D, 使用SQL Server 2012存储):

研究期间,除瀚岳人体行为数据集外,还通过Florence3D Actions(Florence-3D)、UTKinectAction3D(UT-3D)、 MSRAction3D、NTU-RGB+D(仅含单人数据集)和TST V2数据集对相关方法、模型进行验证。为方便使用,将上述所有数据集进行了解析,并以统一格式存储于数据库(近包含关节点的位置信息,不包含元数据集中的图片和视频)。
Other Datasets(include HanYue-3D, use SQL Server 2012 to store):

During the research works, except HanYue-3D dataset, Florence3D Actions(Florence-3D)、UTKinectAction3D(UT-3D)、 MSRAction3D、NTU-RGB+D(only include single person skeleton motion information)and TST V2 were adopted to evaluate my algorithms and models. For ease of use, I analyzed the structures of all the above datasets and stored the joint positions into the SQL Server 2012. The stored motion information only contains the joint positions, does not contain the pictures and videos of the original datasets.
点击此处下载上述解析后的数据集
click here to download above datasets