Performance evaluation of a 3D multi-view-based particle filter for visual object tracking using GPUs and multicore CPUs

by David Concha, Raúl Cabido, Juan José Pantrigo, Antonio S. Montemayor
Abstract:
This paper presents a deep and extensive performance analysis of the particle filter (PF) algorithm for a very compute intensive 3D multi-view visual tracking problem. We compare different implementations and parameter settings of the PF algorithm in a CPU platform taking advantage of the multithreading capabilities of the modern processors and a graphics processing unit (GPU) platform using NVIDIA CUDA computing environment as developing framework. We extend our experimental study to each individual stage of the PF algorithm, and evaluate the quality versus performance trade-off among different ways to design these stages. We have observed that the GPU platform performs better than the multithreaded CPU platform when handling a large number of particles, but we also demonstrate that hybrid CPU/GPU implementations can run almost as fast as only GPU solutions.
Reference:
Performance evaluation of a 3D multi-view-based particle filter for visual object tracking using GPUs and multicore CPUs (David Concha, Raúl Cabido, Juan José Pantrigo, Antonio S. Montemayor), In Journal of Real-Time Image Processing, 2014.
Bibtex Entry:
@Article{Concha2014,
author="Concha, David
and Cabido, Ra{'u}l
and Pantrigo, Juan Jos{'e}
and Montemayor, Antonio S.",
title="Performance evaluation of a 3D multi-view-based particle filter for visual object tracking using GPUs and multicore CPUs",
journal="Journal of Real-Time Image Processing",
year="2014",
pages="1--19",
abstract="This paper presents a deep and extensive performance analysis of the particle filter (PF) algorithm for a very compute intensive 3D multi-view visual tracking problem. We compare different implementations and parameter settings of the PF algorithm in a CPU platform taking advantage of the multithreading capabilities of the modern processors and a graphics processing unit (GPU) platform using NVIDIA CUDA computing environment as developing framework. We extend our experimental study to each individual stage of the PF algorithm, and evaluate the quality versus performance trade-off among different ways to design these stages. We have observed that the GPU platform performs better than the multithreaded CPU platform when handling a large number of particles, but we also demonstrate that hybrid CPU/GPU implementations can run almost as fast as only GPU solutions.",
issn="1861-8219",
doi="10.1007/s11554-014-0483-1",
url="http://dx.doi.org/10.1007/s11554-014-0483-1"
}