A temporal-spatial method for group detection, locating and tracking

Abstract

With the prevalence of smart devices, such as smart phones, wearable equipments, and infrastructures, location-based service (LBS) has thrived in our daily life. In those practical LBS applications, group detection and tracking is a context-related research field in many scenarios, such as school yard, office building, shopping mall and so on. In this paper, we heuristically develop a temporal-spatial method for clustering and locating the groups, and then leverage a CRF-based event detection mechanism to improve the performance of recognizing contextual behaviors. The experimental results demonstrate that our system can achieve an impressive accuracy and precision of grouping and tracking.