• DocumentCode
    596384
  • Title

    GPU-based real-time face detector

  • Author

    Jae-chan Jeong ; Ho-chul Shin ; Jae-il Cho

  • Author_Institution
    IRobot/Cognitive Syst. Res. Dept., Electron. & Telecommun. Res. Inst., Daejeon, South Korea
  • fYear
    2012
  • fDate
    26-28 Nov. 2012
  • Firstpage
    173
  • Lastpage
    175
  • Abstract
    The speed of face detection influences overall performance of applications like face recognition and human-robot interaction (HRI). In some applications such as long distance HRI, face detection is carried out on high resolution image up to 800×600 or more, leading to an increase in detection time and, in turn, deteriorating the application performance. To improve the speed of face detection, we propose a GPU-based face detection system consisting of modified census transform, AdaBoost and four phase cascades. To operate the face detection on GPU, we use CUDA. Experimental results demonstrated that the speed of our system than a CPU-based one was higher by a factor of five to eight compared with a CPU-based face detection system.
  • Keywords
    face recognition; graphics processing units; human-robot interaction; image resolution; object detection; parallel architectures; transforms; AdaBoost; CUDA; GPU-based face detection; face recognition; four phase cascades; high resolution image; human-robot interaction; long distance HRI; modified census transform; real-time face detector; Detectors; Face; Face detection; Graphics processing units; Image resolution; Lighting; Neural networks; CUDA; Face Detection; GPU; Human Robot Interaction;
  • fLanguage
    English
  • Publisher
    ieee
  • Conference_Titel
    Ubiquitous Robots and Ambient Intelligence (URAI), 2012 9th International Conference on
  • Conference_Location
    Daejeon
  • Print_ISBN
    978-1-4673-3111-1
  • Electronic_ISBN
    978-1-4673-3110-4
  • Type

    conf

  • DOI
    10.1109/URAI.2012.6462961
  • Filename
    6462961