GaHu-Video: Parametrization system for human gait recognition
收藏资源简介：
This video dataset comprises (44) recordings of the human gait with (396) edited videos, there is also available a dataset of geometric features with (1431) files acquired from the video dataset that can be used for identifying people by human gait or for analyzing gait on various purposes. Each directory has the following structure: One directory with original videos of 44 people walking. Some parameters are, Video format AVI, stable background, resolution of 1280x700 p. Each video recording registers a person walking from left to right and back three times. One directory (Track A) with 44 edited videos with the first walking pass from left to right and back. A subdirectory (Sx_Track 1_Right) with 44 edited videos with the first walking pass from left to right. A subdirectory (Sx_Track 1_Left) with 44 edited videos with the first walking pass from right to left. This directory comprises 132 videos in total. One directory (Track B) with 44 edited videos with the second walking pass from left to right and back. A subdirectory (Sx_Track 1_Right) with 44 edited videos with the second walking pass from left to right. A subdirectory (Sx_Track 1_Left) with 44 edited videos with the second walking pass from right to left. This directory comprises 132 videos in total. One directory (Track C) with 44 edited videos with the third walking pass from left to right and back. A subdirectory (Sx_Track 1_Right) with 44 edited videos with the third walking pass from left to right. A subdirectory (Sx_Track 1_Left) with 44 edited videos with the third walking pass from right to left. This directory comprises 132 videos in total. One directory V-Geometric features with .dat files acquired from the video dataset with geometric features from the rectangle drawn over the silhouettes during a time. The geometric features were acquired with image processing techniques and are comprised of width, height, and area over time [1,2] nnW subdirectory: 477 .dat files with registered information about the behavior of the rectangle width during the gait record. nnH subdirectory: 477 .dat files with registered information about the behavior of the rectangle height during the gait record. nnA subdirectory: 477 .dat files with registered information about the behavior of the rectangle area during the gait record. References: [1] Senigagliesi, L.; Ciattaglia, G.; De Santis, A.; Gambi, E. People Walking Classification Using Automotive Radar. Electronics 2020, doi:10.3390/electronics9040588 [2] Kececi, Aybuke, Armağan Yildirak, Kaan Ozyazici, Gulsen Ayluctarhan, Onur Agbulut, and Ibrahim Zincir. 2020. Implementation of Machine Learning Algorithms for Gait Recognition. doi:10.1016/j.jestch.2020.01.005 [3] Figueiredo, Joana, Cristina P. Santos, and Juan C. Moreno. 2018. Automatic Recognition of Gait Patterns in Human Motor Disorders using Machine Learning: A Review. Vol. 53. doi: 10.1016/j.medengphy.2017.12.006
本视频数据集包含44段人体步态录制素材，衍生出396条编辑后视频；同时附带从该视频数据集提取得到的1431个几何特征文件数据集，可用于基于人体步态的身份识别或多场景步态分析。每个目录遵循如下结构： 1. 原始视频目录：存储44名行走受试者的原始步态视频，参数包括：视频格式为音频视频交错格式（AVI），背景稳定，分辨率为1280×700像素。每条视频记录一名受试者从左至右单向行走、随后折返，重复该过程共计3次的完整步态。 2. Track A目录：包含44条对应首次往返行走过程的编辑后视频，其下设两个子目录： - Sx_Track 1_Right：包含44条对应首次从左至右单向行走的编辑后视频 - Sx_Track 1_Left：包含44条对应首次从右至左单向行走的编辑后视频 该目录总计包含132条视频。 3. Track B目录：包含44条对应第二次往返行走过程的编辑后视频，其下设两个子目录： - Sx_Track 1_Right：包含44条对应第二次从左至右单向行走的编辑后视频 - Sx_Track 1_Left：包含44条对应第二次从右至左单向行走的编辑后视频 该目录总计包含132条视频。 4. Track C目录：包含44条对应第三次往返行走过程的编辑后视频，其下设两个子目录： - Sx_Track 1_Right：包含44条对应第三次从左至右单向行走的编辑后视频 - Sx_Track 1_Left：包含44条对应第三次从右至左单向行走的编辑后视频 该目录总计包含132条视频。 5. V-Geometric features目录：存储从视频数据集提取得到的.dat格式几何特征文件，此类特征通过图像处理技术提取，为随时间变化的人体轮廓外接矩形相关参数，包括宽度、高度与面积[1,2]。该目录下设三个子目录： - nnW：包含477个.dat文件，记录步态录制过程中外接矩形宽度的变化信息 - nnH：包含477个.dat文件，记录步态录制过程中外接矩形高度的变化信息 - nnA：包含477个.dat文件，记录步态录制过程中外接矩形面积的变化信息 参考文献： [1] Senigagliesi, L.; Ciattaglia, G.; De Santis, A.; Gambi, E. People Walking Classification Using Automotive Radar. Electronics 2020, doi:10.3390/electronics9040588 [2] Kececi, Aybuke, Armağan Yildirak, Kaan Ozyazici, Gulsen Ayluctarhan, Onur Agbulut, and Ibrahim Zincir. 2020. Implementation of Machine Learning Algorithms for Gait Recognition. doi:10.1016/j.jestch.2020.01.005 [3] Figueiredo, Joana, Cristina P. Santos, and Juan C. Moreno. 2018. Automatic Recognition of Gait Patterns in Human Motor Disorders using Machine Learning: A Review. Vol. 53. doi: 10.1016/j.medengphy.2017.12.006
