Muslim Prayer Actions Recognition

Action recognition is one of the most interesting research areas in the last decade. A wide variety of models were proposed in the context of developing human action recognition systems. In this paper we focus on Muslim prayer actions recognition. We present a system capable of recognizing major postures and actions performed by a Muslim normally during the prayer, an unexplored area in the action recognition field. The paper presents the details of the system implementation phases and the proposed techniques used for detection. The system was tested against a custom data set and an accuracy exceeding 93.5% was recorded. The paper is concluded by the challenges and thoughts concerning future extensions of this work.