A real-time human identification system through multimodal information

摘要

Human identification is very important for many applications. Most existing methods are based on ID cards or biometrical methods, such as fingerprint hard information or clothing color soft biometric trait. In this work, we propose a novel framework for human identification through recognizing human moving trajectory by two modalities, that is, the motion information from bird's eye view depth cameras and the compass sensor signals of a mobile device. Our proposed system has the advantages of ease of installation and support real-time service. We believe the proposed system will have a significant promise for human identification service.