You are here

Stefano Messelodi

Head of Unit
  • Phone: 0461314509
  • FBK Povo
Short bio

Stefano Messelodi was born in Arco (Italy) in 1961. He graduated in computer science from the University of Milan (Italy).  Since 1986 he is working in FBK (ITC-irst), Trento, Italy, where he coordinated the Technologies of Vision research unit till 2012. His research interests include text localization in scene, semantic image labelling and dynamic scene understanding. He served as a reviewer for several journals and conferences. He is a member of IEEE Society and International Association for Pattern Recognition.
He currently coordinates the research unit TeV.

Research interests
people re-identification text localization in scene document image analysis traffic analysis
Publications
  1. G. Zen; O. Lanz; S. Messelodi; E. Ricci,
    Tracking Multiple People with Illumination Maps,
    Proceedings of International Conference on Pattern Recognition,
    IEEE,
    2010
    , pp. 3484-
    3487
    , (International Conference on Pattern Recognition (ICPR),
    Istanbul, Turkey,
    08/23/2010 - 08/26/2010)
  2. A. Pnevmatikakis; N. Katsarakis; P. Chippendale; C. Andreatta; S. Messelodi; C.M. Modena; F. Tobia,
    Proceedings of the 2010 ACM workshop on Social, adaptive and personalized multimedia interaction and access - SAPMIA 2010,
    New York, NY,
    ACM,
    2010
    , pp. 67-
    72
    , (ACMMM International Multimedia Conference,
    Firenze, Italy,
    from 10/29/2010 to 10/29/2010)
  3. O. Lanz; S. Messelodi,
    Method for efficient target detection from images robust to occlusion.
    2010,
    The method for efficient target detection from images robust to occlusion disclosed by the present invention detects the presence and spatial location of a number of objects in images. It consists in (i) an off-line method to compile an intermediate representation of detection probability maps that are then used by (ii) an on-line method to construct a detection probability map suitable for detecting and localizing objects in a set of input images efficiently. The method explicitly handles occlusions among the objects to be detected and localized, and objects whose shape and configuration is provided externally, for example from an object tracker. The method according to the present invention can be applied to a variety of objects and applications by customizing the method's input functions, namely the object representation, the geometric object model, its image projection method, and the feature matching function.
  4. Michela Lecca; Stefano Messelodi,
    Planckian Illuminants and Von Kries Model,
    Technical Report -

    In this work the authors investigate two hypotheses commonly assumed in Computer Vision and Computer Graphics for solvin

    In this work the authors investigate the relations between the colors of two pictures of a same scene taken under two different light sources. This is a crucial task in Computer Vision and Computer Graphics.
    The authors show that: (1) the von Kries model suffices for describing the color variation due to a Planckian illuminant change; (2) the Planck's law constraints the triplets of the von Kries coefficients to be points of a ruled surfaces parametrized by the photometric cues of the illuminants; (3) the von Kries coefficients are strongly related to the sensitivity functions of the device.
    Several experiments carried out on public image datasets are presented.

    ,
    2010
  5. C. M. Modena; S. Messelodi,
    Recognition and tracking of embedded text in sport videos: a case study,
    Abstract - We present an athlete recognition module designed for broadcast videos, forming part of a system designed for the personalization of sport video broadcast. The aim of this module is the identification of athletes in the scene through the reading of names or numbers printed on their uniforms and to identify frames where athletes are visible. Using an adaptation of a previously published algorithm we extract text from individual frames and then read these candidates by means of an optical character recognizer (OCR). The OCR-ed text is then compared to an a-priori list of athletes' names (or numbers), to provide a presence score for each athlete. Text regions are subsequently tracked in following frames using a template matching technique. In this way blurred or distorted text, normally unreadable by the OCR, can also be exploited to provide a denser labelling of the video sequences.,
    2010
  6. Stefano Messelodi; Carla Maria Modena; Michele Zanin; Francesco De Natale; Fabrizio Granelli; Enrico Betterle; Andrea Guarise,
    in «EXPERT SYSTEMS WITH APPLICATIONS»,
    vol. 36,
    n. 3 Part 1,
    2009
    , pp. 4213 -
    4227
  7. Michela Lecca; Stefano Messelodi,
    Computing von Kries Illuminant Changes by Piecewise Inversion of Cumulative Color Histograms,
    in «ELCVIA. ELECTRONIC LETTERS ON COMPUTER VISION AND IMAGE ANALYSIS»,
    2009
    , pp. 1 -
    17
  8. Michela Lecca; Stefano Messelodi,
    Illuminant Change Estimation via Minimization of Color Histogram Divergence,
    Computational Color Imaging Workshop,
    Springer,
    vol.5646/2009,
    n. LNCS 5646,
    2009
    , pp. 41-
    50
    , (Computational Color Imaging Workshop,
    Saint-Etienne, France,
    03/26/2009 - 03/27/2009)
  9. O. Lanz; S. Messelodi,
    A sampling algorithm for occlusion robust multi target detection,
    2009
    , pp. 346-
    351
    , (6th IEEE International Conference on Advanced Video and Signal based Surveillance - AVSS 2009,
    Genova, Italy,
    da 09/02/2009 a 09/04/2009)
  10. O. Lanz; S. Messelodi,
    Method for efficient target detection from images robust to occlusion.
    2009,
    The method for efficient target detection from images robust to occlusion disclosed by the present invention detects the presence and spatial location of a number of objects in images. It consists in (i) an off-line method to compile an intermediate representation of detection probability maps that are then used by (ii) an on-line method to construct a detection probability map suitable for detecting and localizing objects in a set of input images efficiently. The method explicitly handles occlusions among the objects to be detected and localized, and objects whose shape and configuration is provided externally, for example from an object tracker. The method according to the present invention can be applied to a variety of objects and applications by customizing the method's input functions, namely the object representation, the geometric object model, its image projection method, and the feature matching function.

Pages