Estimation of Interests and Classification of Behavior Patterns with Trajectory Analysis Used for Event Sites

Abstract

This paper presents a method for classification and recognition of behavior patterns based on interests from human trajectories at an event site. Our method creates models of Hidden Markov Models (HMMs) for each human trajectory quantized using one-dimensional Self- Organizing Maps (SOMs). Subsequently, we apply two-dimensional SOMs for unsupervised classification of behavior patterns from features of the distance between models. Furthermore, we use Unified distance Matrix (U-Matrix) for visualizing category boundaries based on the Euclidean distance between weights of SOMs. Our method can extract typical behavior patterns and peculiar behavior patterns of interests obtained using a questionnaire and visualize relationships between these patterns. We evaluated our method based on Cross Validation (CV) methods using trajectories of typical behavior patterns. The recognition accuracy reached to 83.3% for estimating interests of three types. We consider that our method is useful to estimate interests from behavior patterns at an event site.