Faculty of Information Technology, BUT

Publication Details

Cascaded Stripe Memory Engines for Multi-Scale Object Detection in FPGA

MUSIL Petr, JURÁNEK Roman, MUSIL Martin and ZEMČÍK Pavel. Cascaded Stripe Memory Engines for Multi-Scale Object Detection in FPGA. IEEE Transactions on Circuits and Systems for Video Technology, vol. 2019, no. 1, pp. 1-13. ISSN 1051-8215. Available from: https://ieeexplore.ieee.org/document/8573854
Czech title
Kaskádová paměťová architektura pro detekci objektů různé velikosti ve videu v FPGA
Type
journal article
Language
english
Authors
Musil Petr, Ing. (DCGM FIT BUT)
Juránek Roman, Ing., Ph.D. (DCGM FIT BUT)
Musil Martin, Ing. (DCGM FIT BUT)
Zemčík Pavel, prof. Dr. Ing. (DCGM FIT BUT)
URL
Keywords
Object Detection, AdaBoost, WaldBoost, Acceleration, FPGA
Abstract
Object detection in embedded systems is important for many contemporary applications that involve vision and scene analysis. In this paper, we propose a novel architecture for object detection implemented in FPGA, based on the Stripe Memory Engine (SME), and point out shortcomings of existing architectures. SME processes a stream of image data so that it stores a narrow stripe of the input image and its scaled versions and uses a detector unit which is efficiently pipelined across multiple image positions within the SME. We show how to process images with up to 4K resolution at high framerates using cascades of SMEs. As a detector algorithm, the SMEs use boosted soft cascade with simple image features that require only pixel comparisons and look-up tables; therefore, they are well suitable for hardware implemenation. We describe the components of our architecture and compare it to several published works in several configurations. As an example, we implemented face detection and license plate detection applications that work with HD images (1280×720 pixels) running at over 60 frames per second on Xilinx Zynq platform. We analyzed their power consumption, evaluated the accuracy of our detectors, and compared them to Haar Cascades from OpenCV that are often used by other authors. We show that our detectors offer better accuracy as well as performance at lower power consumption.
Published
2019 (in print)
Pages
1-13
Journal
IEEE Transactions on Circuits and Systems for Video Technology, vol. 2019, no. 1, ISSN 1051-8215
Publisher
IEEE Circuits and Systems Society
DOI
BibTeX
@ARTICLE{FITPUB11896,
   author = "Petr Musil and Roman Jur\'{a}nek and Martin Musil and Pavel Zem\v{c}\'{i}k",
   title = "Cascaded Stripe Memory Engines for Multi-Scale Object Detection in FPGA",
   pages = "1--13",
   journal = "IEEE Transactions on Circuits and Systems for Video Technology",
   volume = 2019,
   number = 1,
   year = 2019,
   ISSN = "1051-8215",
   doi = "10.1109/TCSVT.2018.2886476",
   language = "english",
   url = "https://www.fit.vut.cz/research/publication/11896"
}
Back to top