How to Solve Common Human Actions Recognition Use Cases with Gesture Description Language

摘要

The main novelty of this paper is explanation of several common use cases that often appears in typical action recognition scenarios. The choice of those scenarios is based on my own practice in action recognition and they also show in my opinion the most effective way of using Gesture Description Language (GDL) for various tasks of action recognition. All resources that were used to prepare experiments described in this paper: motion capture data, GDL script codes and GDL Studio (an application for motion capture, editing and classification) are open sources. Due to this the whole solution might be easily applied to scientific experiments and industrial applications. This paper also explains the important issues of GDL theory and practice.