Fast Newton active appearance models

Jean Kossaifi, Georgios Tzimiropoulos, Maja Pantic

    Research output: Chapter in Book/Report/Conference proceedingConference contributionAcademicpeer-review

    6 Citations (Scopus)
    15 Downloads (Pure)

    Abstract

    Active Appearance Models (AAMs) are statistical models of shape and appearance widely used in computer vision to detect landmarks on objects like faces. Fitting an AAM to a new image can be formulated as a non-linear least-squares problem which is typically solved using iterative methods. Owing to its efficiency, Gauss-Newton optimization has been the standard choice over more sophisticated approaches like Newton. In this paper, we show that the AAM problem has structure which can be used to solve efficiently the original Newton problem without any approximations. We then make connections to the original Gauss-Newton algorithm and study experimentally the effect of the additional terms introduced by the Newton formulation on both fitting accuracy and convergence. Based on our derivations, we also propose a combined Newton and Gauss-Newton method which achieves promising fitting and convergence performance. Our findings are validated on two challenging in-the-wild data sets.
    Original languageUndefined
    Title of host publicationProceedings of IEEE International Conference on Image Processing (ICIP 2014)
    Place of PublicationUSA
    PublisherIEEE Computer Society
    Pages1420-1424
    Number of pages5
    ISBN (Print)978-1-4799-5751-4
    DOIs
    Publication statusPublished - Oct 2014
    EventIEEE International Conference on Image Processing, ICIP 2014 - Paris, France
    Duration: 27 Oct 201430 Oct 2014
    https://icip2014.wp.imt.fr/

    Publication series

    Name
    PublisherIEEE Computer Society

    Conference

    ConferenceIEEE International Conference on Image Processing, ICIP 2014
    Abbreviated titleICIP
    CountryFrance
    CityParis
    Period27/10/1430/10/14
    Internet address

    Keywords

    • HMI-HF: Human Factors
    • inverse compositional imagealignment
    • EWI-25825
    • EC Grant Agreement nr.: FP7/2007-2013
    • EC Grant Agreement nr.: FP7/611153
    • METIS-309950
    • LevenbergMarquardt
    • Newton method
    • IR-95231
    • Active Appearance Models
    • EC Grant Agreement nr.: FP7/288235

    Cite this

    Kossaifi, J., Tzimiropoulos, G., & Pantic, M. (2014). Fast Newton active appearance models. In Proceedings of IEEE International Conference on Image Processing (ICIP 2014) (pp. 1420-1424). USA: IEEE Computer Society. https://doi.org/10.1109/ICIP.2014.7025284