[visionlist] Postdoc position 2008 in Content-Based Image Retrieval: Structuring local description of visual contents - CEDRIC, Paris France

[Valerie Gouet-Brunet]

*Post-doctoral position 2008
*
*Title:* Structuring local description of visual contents
*Keywords: *Content-Based Image/Video Retrieval, Local and semi-local 
image descriptors, Psychovisual assessment, Object recognition, 
Content-Based Copy Detection.**
*Duration:*1 year from January 1^st 2008, renewable 1 year

*Research groups and labs:
***This post-doc will be conducted in the French Laboratory in Computer 
Vision CEDRIC (Conservatoire National des Arts et Métiers) in Paris, in 
collaboration with the SIC Laboratory of the University of Poitiers. The 
recruited researcher will work within CEDRIC/Vertigo and SIC/ICONES 
research groups.
CEDRIC: http://cedric.cnam.fr/index.php?lang=en
CEDRIC/Vertigo: http://cedric.cnam.fr/AfficheEquipe.php?id=9&lang=en 
<http://cedric.cnam.fr/AfficheEquipe.php?id=9&lang=en>
SIC: http://www.sic.sp2mi.univ-poitiers.fr/index.php
SIC/ICONES: http://www.sic.sp2mi.univ-poitiers.fr/themes/icones/
 
*Location:*
CEDRIC is located in the center of Paris and SIC in the Futuroscope 
area, near the city of Poitiers. Some travels will be necessary between 
these locations.
 
*Description:*
This post-doc deals with content-based indexing and retrieval in image 
and video databases (CBIR). We will focus on methods involving /local 
and semi-local description/ of the image content. To improve 
description, some approaches combine several natures of visual features, 
and they often push forward the homogeneity or discontinuity of the 
considered patterns. However, most of the time, these methods only 
combine two types of features, without any strong justification about 
the complementarity and optimality of such associations. The work 
concerns the study of a description of the visual content of images and 
videos, from local and semi-local descriptors that could be enriched by 
investigating the two orthogonal axes of research described below:
* *
*1) Combination of heterogeneous features.* The first objective will 
concern the evaluation of the complementarity of the local and 
semi-local descriptors found in literature. Criteria on the relevance of 
these features and their combination will be defined according to 
considered applications. In particular, we will focus on their 
/psychovisual concordance/, by conducting eye tracking experiments, 
within the psychovisual facilities of the SIC. We think that such a 
study will allow to bring to light synergies between relevant 
descriptors, while maintaining their compacity.**
* *
*2) Spatial structuring of local features. *Also with the aim to improve 
local or semi-local description of images, it is possible to consider 
their spatial organization in the image. Such an information highly 
contributes to enrich the description, but at the cost of a 
non-negligible complexity during matching. In this work, we propose to 
study a hierarchical structuring of these relations, and to introduce 
the context of psychovisual perception during the matching process.
 
During this post-doc, we will consider the following applications: 
object recognition in images and videos for videosurveillance, and 
content-based copy detection. Local and semi-local descriptors are 
particularly useful for these applications.
 
*Required skills:*
- PhD in content-based image or video retrieval, or in image analysis or 
computer vision.
- Very good knowledge on image or video descriptors.
- Experience in the problem of scalability (curse of dimensionality).
* *
*Contacts:*
- Valérie Gouet-Brunet (CEDRIC/Vertigo) : valerie.gouet at cnam.fr 
<mailto:valerie.gouet at cnam.fr>
- Noël Richard (SIC/ICONES) : richard at sic.sp2mi.univ-poitiers.fr 
<mailto:richard at sic.sp2mi.univ-poitiers.fr>
- Christine Fernandez-Maloigne (SIC/ICONES) : 
fernandez at sic.sp2mi.univ-poitiers.fr 
<mailto:fernandez at sic.sp2mi.univ-poitiers.fr>
* *
*Application procedure:*
Send your application by email to the three contacts, *before end of 
december 2007*: a detailed CV, a letter explaining your motivations for 
the topic and recommendation letters from three referees.

-- 
______________________________________________________________

Valerie Gouet-Brunet
CNAM - CC 432               INRIA Rocquencourt - Imedia Group
292, rue St. Martin         Domaine de Voluceau, BP 105
F75141 Paris Cedex 03       F78153 Le Chesnay Cedex
Tel : +33 1 58 80 86 35     Tel : +33 1 39 63 57 81
Fax : +33 1 58 80 84 93     Fax : +33 1 39 63 56 74
______________________________________________________________ 

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://visionscience.com/pipermail/visionlist/attachments/20071103/6cff4400/attachment.htm