Towards tool support for computational causal behavior models for activity recognition

Abstract

Complexity in Modeling and Development increases continuously. Subject of this paper is to show how developers of Computational Causal Behavior Models can be supported with a tool to improve the development process. Therefore, we analyze three case studies about probabilistic modeling of human behavior for activity recognition with respect to specific problems that might occur. We present a classification into five problem types that should be carefully considered in such projects. Finally, we present how tool support to increase the efficiency of developing Computational Causal Behavior Models might be created. In general, the paper provides useful guidelines for every model developer in the field of activity recognition, and highlights the need of tool support for such models.