Dataset for Human Recognition under Multi-Gait Scenario

Previous research in gait recognition and available dataset focused on single subjects. But in real time environment (such as airport, metro stations etc.) more than one person walk in a group and occlusion factors affects the performance of gait recognition. Therefore, we presented a new dataset which focused dynamic occlusion scenario. We constructed two categories of gait dataset i.e. first, subjects walk in a group (SMVDU-Multi-Gait) and second, subjects walk separately (SMVDU-Single-Gait). The purpose of this dataset to analyze the variations in gait patterns when same subject walk in a group or individually and also to recognize the target subject in Multi-Gait.