BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Date iCal//NONSGML kigkonsult.se iCalcreator 2.20.2//
METHOD:PUBLISH
X-WR-CALNAME;VALUE=TEXT:Eventi DIAG
BEGIN:VTIMEZONE
TZID:Europe/Paris
BEGIN:STANDARD
DTSTART:20171029T030000
TZOFFSETFROM:+0200
TZOFFSETTO:+0100
TZNAME:CET
END:STANDARD
BEGIN:DAYLIGHT
DTSTART:20170326T020000
TZOFFSETFROM:+0100
TZOFFSETTO:+0200
RDATE:20180325T020000
TZNAME:CEST
END:DAYLIGHT
END:VTIMEZONE
BEGIN:VEVENT
UID:calendar.12453.field_data.0@www.ugov-ricerca.uniroma1.it
DTSTAMP:20260408T063212Z
CREATED:20170922T084538Z
DESCRIPTION:Figure-ground separation is a land-mark problem in visual perce
 ption\, which has fascinated many scientists for centuries. In computer vi
 sion\, edge detection and region segmentation of an image have been two gr
 and challenges for understanding of an image in terms of objects and conte
 xtual surroundings\, and shapes and appearances of objects.  Generic Segme
 ntation of an image involves grouping pixels\, which are perceptually simi
 lar. However\, in Semantic Segmentation the aim is to assign a semantic la
 bel to each pixel in the image. Even though semantic segmentation can be a
 chieved by simply applying classifiers (which are trained using supervised
  learning)\, to each pixel or a region in the image\, the results may not 
 be desirable due to the fact that general context information beyond the s
 imple smoothness is not considered. In this talk\, I will start with brief
 ly presenting two supervised approaches to address this problem.  First\, 
 I will discuss an approach to discover interactions between labels and reg
 ions using a sparse estimation of precision matrix\, which is the inverse 
 of covariance matrix of data obtained by graphical lasso. In this context\
 , we find a graph over labels as well as regions in the image which encode
 s significant interactions and also it is able to capture the long-distanc
 e associations. Second\, I will introduce a knowledge-based method to inco
 rporate dependencies among regions in the image during inference. High lev
 el knowledge rules - such as co-occurrence\, spatial relations and mutual 
 exclusivity - are extracted from training data and transformed into constr
 aints in Integer Programming formulation. A difficulty which most supervis
 ed semantic segmentation approaches are confronted with is lack of enough 
 training data\, particularly in deep learning methods which have become en
 ormously popular recently. Annotated data should be at the pixel-level (i.
 e.\, each pixel of training images must be annotated)\, which is highly ex
 pensive to achieve. To address this limitation\, next I will present a sem
 i supervised learning approach to exploit the plentiful amount of availabl
 e unlabeled as well as synthetic images generated via Generative Adversari
 al Networks (GAN). Furthermore\, I will discuss an extension of the model 
 to use additional weakly labeled data to solve the problem in a weakly sup
 ervised manner. The basic idea here is by providing these fake data from t
 he Generator and the competition between real/fake data (discriminator/gen
 erator networks)\, true samples are encouraged to be close in the feature 
 space. Therefore\, the model learns more discriminative features\, which l
 ead to better classification results for semantic segmentation. Biographic
 al Sketch:Dr. Mubarak Shah\, the UCF Trustee Chair Professor\, is the foun
 ding director of Center for Research in Computer Visions at University of 
 Central Florida (UCF). He is a co-author of five books (Motion-Based Recog
 nition (1997)\; Video Registration (2003)\; Automated Multi-Camera Surveil
 lance: Algorithms and Practice (2008)\; Modeling\, Simulation and Visual A
 nalysis of Crowds (2013)\; and Robust Subspace Estimation Using Low-Rank O
 ptimization (2014)\, all by Springer.  He has published extensively on top
 ics related to visual surveillance\, tracking\, human activity and action 
 recognition\, object detection and categorization\, shape from shading\, g
 eo registration\, visual crowd analysis\, etc. Dr. Shah is a fellow of IEE
 E\, IAPR\, AAAS and SPIE.  He has been ACM and IEEE Distinguished Visitor 
 Program speaker and is often invited to present seminars\, tutorials and i
 nvited talks all over the world. He received Pegasus award in 2006\; Unive
 rsity Distinguished Research Award in 2017\, 2012 and 2005\; Faculty Excel
 lence in Mentoring Doctoral Students in 2016\, Scholarship of Teaching and
  Learning award in 2011\; Teaching Incentive Program award in 1995 and 200
 3\; Research Incentive Award in 2003\, 2009 and 2012\; the Harris Corporat
 ion Engineering Achievement Award in 1999\; the TOKTEN awards from UNDP in
  1995\, 1997\, and 2000\; 2009 IEEE Outstanding Engineering Educator Award
  in 1997\; an honorable mention for the ICCV 2005 Where Am I? Challenge Pr
 oblem\; 2013 NGA Best Research Poster Presentation\, 2nd place in Grand Ch
 allenge at the ACM Multimedia 2013 conference\; and runner up for the best
  paper award in ACM Multimedia Conference in 2005 and 2010.
DTSTART;TZID=Europe/Paris:20171016T140000
DTEND;TZID=Europe/Paris:20171016T140000
LAST-MODIFIED:20191008T082902Z
LOCATION:Aula MAgna
SUMMARY:Solving Semantic Segmentation: Precision Matrix\, Knowledge-Based R
 ules and Generator Adversarial Network (GAN) - Prof. Mubarak Shah
URL;TYPE=URI:http://www.ugov-ricerca.uniroma1.it/node/12453
END:VEVENT
END:VCALENDAR