Virtualized reality model-based benchmarking of AR/MR camera tracking methods in TrakMark

We describe a virtualized reality model-based benchmarking of camera tracking methods for augmented reality and mixed reality. Benchmark datasets created with virtualized reality models do not include any measurement errors, and can be effectively used for a benchmark of tracking methods [5][6]. But there are many issues to be discussed for creators and users of the datasets. For the creators, efficient creation and sharing of the datasets are priority issues. On the other hand, supports for benchmarking processes and for creating original datasets are desired by the users. To tackle these issues, we propose a prototype framework for a virtualized reality modelbased benchmarking of camera tracking methods. In this paper, first, we describe an outline of the framework, and show actual components in the framework. Moreover, we show experimental results of benchmarking our tracking method with the framework.