Real time Multi-GPU-based Event Detection in High Definition Videos
University of Mons, Faculty of Engineering, Computer Science Department, Place du Parc, 20, 7000 Mons, Belgium
5th International Conference on Electronics Engineering (ICEE 2013), 2013
@article{mahmoudi2013real,
title={Real time Multi-GPU-based Event Detection in High Definition Videos},
author={Mahmoudi, Sidi Ahmed and Manneback, Pierre},
year={2013}
}
Video processing algorithms present a very important tool for many applications related to computer vision domain such as motion tracking, videos indexation, robot navigation and event detection. However, the new video standards, especially in high definitions, cause that the current implementations, even running on modern hardware, no longer respect the needs of real-time processing. In this context, several solutions have been proposed to overcome this constraint, by exploiting graphic processing units (GPUs). Although, they present a high potential of GPU, any is able to treat high definition videos efficiently. This paper presents a real time method that enables to detect portions of video that correspond to sudden changes of motion variations of movements. Experimental results have been conducted using several videos showing several events that have been detected within different scenarios. The simultaneous exploitation of multiple GPUs enabled a real time treatment of high definition videos with a global speedup ranging from 5 to 33, by comparison with CPU implementations.
December 15, 2013 by hgpu
