image1 image2 image3

Papers & Publications

Jun 14, 2012

Evaluation of Visual Descriptors for Multi-camera Object Identification in Surveillance Systems


(Under review)

"Pattern Analysis and Applications" Journal (Springer) - Damian Ellwart (Politechnika Gdanska, Poland), Piotr Dalka (Politechnika Gdanska, Poland) and Andrzej Czyzewski (Politechnika Gdanska, Poland).

Comparative analysis of various visual descriptors is presented. The descriptors utilize various aspects of image data: color, texture, gradient and statistical moments. The descriptor list is supplemented with local features calculated in local vicinities of keypoints found automatically in the image.

The goal of the analysis is to find descriptors that are best suited for identification of objects in a multi-camera environment. The analysis is performed using two datasets containing images of humans and vehicles recorded with various cameras.

For the purpose of descriptor evaluation, two scatter and clustering measures are supplemented with the new measure that is based on calculating direct dis-similarities between pairs of images. In order to draw conclusions from multi-dataset analysis, four aggregation measures are introduced. They are meant to find descriptors that provide the best identification effectiveness based on the relative ranking and that are simultaneously characterized with large stability (invariance to the selection of objects in the dataset). The results achieved are discussed in details and illustrated with figures. Visual descriptors best suited for identification of humans, vehicles and both object types are proposed.



< Return |  Print

ADDPRIV presentation


Latest News

Feb 20, 2014

New testing session for the ADDPRIV consortium in Milan Linate airport

The consortium partners are further testing the privacy sensitive technology of ADDPRIV in order to increase its stability in real environment

Jan 13, 2014

ADDPRIV prototype is evaluated in Milan-Linate Airport

ADDPRIV Consortium will meet in Milan Linate Airport to test the ADDPRIV system.




Copyright © 2011 ADDPRIV

The research leading to these results has received funding from the European Community's Seventh Framework Programme [FP7/2007-2013] under grant agreement n° 261653

This website reflects the views only of the ADDPRIV Consortium, and the Commission cannot be held responsible for any use which may be made of the information contained herein