The UAV person tracking task for this evaluation was particularly difficult because of large, complicated, and low-quality videos, with only small images of people. We found that our best results were obtained using a combination of intensity thresholding (for IR imagery), motion compensation, interest-point detection and correspondence, and pattern classification. This can be considered a preliminary exploration into an extremely challenging problem.


Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.


  1. 1.
    Ali, S., Shah, M.: Cocoa - tracking in aerial imagery. In: SPIE Airborne Intelligence, Surveillance, Reconnaissance (ISR) Systems and Applications, Orlando (2006)Google Scholar
  2. 2.
    Javed, O., Shah, M.: Tracking and object classification for automated surveillance. In: The Seventh European Conference on Computer Vision, Denmark (2002)Google Scholar
  3. 3.
    Sheikh, Y., Zhai, Y., Shah, M.: An accumulative framework for alignment of an image sequence. In: Proceedings of Asian Conference on Computer Vision (2004)Google Scholar
  4. 4.
    Shafique, K., Shah, M.: A noniterative greedy algorithm for multiframe point correspondence. In: IEEE Trans. Pattern Anal. Mach. Intell. (2005)Google Scholar
  5. 5.
    Ahmed, J., Jafri, M.N., Shah, M., Akbar, M.: Real-time edge-enhanced dynamic correlation and predictive open-loop car-following control for robust tracking. Machine Vision and Applications Journal, Manuscript submission ID MVA-May-06-0110 (accepted, 2007)Google Scholar

Copyright information

© Springer-Verlag Berlin Heidelberg 2008

Authors and Affiliations

  • Andrew Miller
    • 1
  • Pavel Babenko
    • 1
  • Min Hu
    • 1
  • Mubarak Shah
    • 1
  1. 1.Computer Vision Lab at University of Central Florida 

Personalised recommendations