零售场景多模态行人步态分析数据集Gait3D

该数据集主要面向零售场景下多模态行人步态分析研究、采集自北京市通州区经济开发区科创十一街7FRESH七鲜生鲜超市(亦城景园店)，采集时间为2020年9月8日至2020年9月16日，采集设备为39个1920×1080分辨率的高清监控摄像头，最终制作而成的数据总量约132 GB，总计包含有4000个行人和25309个行人步态序列。其中划分训练集为3000个行人、18940个行人序列；查询集为1000个行人的1000个行人序列；匹配集为1000个行人的5369个行人序列。该数据集经过脱敏处理，不包含任何个人信息和敏感信息。This dataset is primarily intended for research on multimodal pedestrian gait analysis in retail scenarios. It was collected at the 7FRESH (SevenFresh) Fresh Supermarket (Yichengjingyuan Branch), No. 11 Chuangke 11th Street, Economic Development Zone, Tongzhou District, Beijing, between September 8 and September 16, 2020. A total of 39 high-definition surveillance cameras with a resolution of 1920×1080 were utilized for data acquisition. The finalized dataset has a total size of approximately 132 GB, encompassing 4000 unique pedestrians and 25309 pedestrian gait sequences in total. The dataset is split into three subsets: the training set comprises 3000 unique pedestrians with 18940 gait sequences; the query set contains 1000 gait sequences from 1000 unique pedestrians; the matching set consists of 5369 gait sequences from 1000 unique pedestrians. This dataset has undergone de-identification processing, and no personally identifiable information or sensitive data is included within it.

搜集汇总

背景与挑战

背景概述

零售场景多模态行人步态分析数据集Gait3D是一个面向零售场景下多模态行人步态分析研究的数据集，采集自北京市通州区经济开发区的一家生鲜超市，使用39个高清监控摄像头采集，数据总量约132GB，包含4000个行人和25309个行人步态序列，分为训练集、查询集和匹配集。数据集经过脱敏处理，不包含个人信息和敏感信息。