CASIA-B 和 PsyMo
收藏资源简介:
CASIA-B和PsyMo数据集是用于步态分析的基准数据集,CASIA-B包含124个不同个体的步态序列,从11个角度记录,而PsyMo包含312个个体的步态视频,从6个角度记录7种不同的场景。这些数据集在受控的实验室环境中收集,旨在研究步态与个体身份识别之间的关系,以及步态如何反映个体的心理特征。数据集的创建过程包括对参与者的步态进行多角度、多场景的录制,并通过预训练的步态识别模型提取步态嵌入,用于后续的步态识别和注册任务。这些数据集广泛应用于步态识别研究中,旨在解决开放集步态注册问题,即确定新的步态样本是否属于数据库中已知的身份,还是代表一个之前未见的个体。
CASIA-B and PsyMo are benchmark datasets for gait analysis. CASIA-B contains gait sequences of 124 distinct individuals, recorded from 11 viewpoints, while PsyMo includes gait videos of 312 individuals, captured across 6 viewpoints and 7 different scenarios. Both datasets are collected in controlled laboratory environments, aiming to investigate the relationship between gait and individual identity recognition, as well as how gait reflects an individual’s psychological characteristics. The dataset creation process involves multi-angle and multi-scenario recording of participants’ gaits, with gait embeddings extracted via pre-trained gait recognition models for subsequent gait recognition and enrollment tasks. These datasets are widely used in gait recognition research, targeting the open-set gait enrollment problem: determining whether a new gait sample belongs to a known identity in the database or represents a previously unseen individual.




