An efficient method for real-time activity recognition

Real-time feature extraction is a key component for any action recognition system that claims to be truly real-time. In this paper we present a conceptually simple and computationally efficient method for real-time human activity recognition based on simple statistical features. Such features are very cheap to compute and form a relatively low dimensional feature space in which classification can be carried out robustly. On the Weizmann dataset, the proposed method achieves encouraging recognition results with an average rate up to 97.8%. These results are in a good agreement with the literature. Further, the method achieves real-time performance, and thus can offer timing guarantees to real-time applications.