Long Period Re-identification Approach to Improving the Quality of Education: A Preliminary Study

Early school leaving is one of the most frequently mentioned reasons to social exclusion later in life. In order to reduce the risk of early school leaving, it is necessary to automate the process of entering unjustified lessons’ delays in school management system. A person’s re-identifying (Re-ID) is a complex automated process, where most studies use an approach to analyze the descriptors of clothing and appearance that are intended for the use of short-period Re-ID. In contrast, there is not much research in the real-time long-term Re-ID process, when images or videos are taken at intervals of several days or months in an uncontrolled environment. In this case descriptors characterizing a person’s biometric identity based on unique features, such as a facial digital image, are required. The objective of this research is to develop a real-time person’s long-term re-identification approach for accounting of non-attended lessons in educational institutions. The proposed Re-ID mechanism includes face identification and new method of using multiple face etalon versions and multiple versions of descriptors for a single person. This allows Re-ID of a person in different clothing and appearance from different camera angles in a long term.