IEEE 2005 Winter Vision Multi-Meeting

IEEE International Workshop on Performance Evaluation of Tracking and Surveillance (PETS)

January 7, 2005

(in conjunction with WACV'05 and Motion'05)

Breckenridge, Colorado

The growth in the development of the field has not been met with complementary systematic performance evaluation of developed techniques. It is especially difficult to make comparisons between algorithms if they have been tested on different datasets under widely varying conditions. 

PETS2005 continues the theme of the highly successful PETS workshops on Performance Evaluation of Tracking and Surveillance. The workshop series is unique in that all participants are testing algorithms on the same datasets. In January 2005, PETS will be co-located along with two other long-running workshops, Motion 2005 and WACV 2005. 

For PETS2005, the theme is target detection and tracking.

The datasets will be made available to participants of the workshop session. The aims of the workshop are several-fold: 

• to bring together researchers interested in the area of target detection and tracking
• to apply different algorithms to the same dataset(s)
• to evaluate the differences between different models and algorithms
• to discuss which criteria should be used for objective evaluation
• to discuss how to document the performance (including accuracy) of visual surveillance algorithms
• to discuss the development of a methodology for testing algorithms
• to discuss the ongoing development of a testbed and guidelines for performing empirical comparative tests.

Travel to Breckenridge

For flights via London, the following airline/travel websites provide good fares:

• British Airways
• Ebookers
• Opodo

For an idea of the proceedings, you may like to consult the electronic proceedings for previous PETS workshops, for example, PETS'2001 located here.

The new data is available here. If you wish to use previously published PETS datasets to test surveillance algorithms, you may wish to consider the PETS repository located here. 

You can access the data via http or via ftp:
ftp> ftp.pets.rdg.ac.uk  with login: anonymous password: your email address.

The datasets include the following:

• [1] outdoor people and vehicle tracking (single camera), available here.  Details on the dataset specification is available here. 

• [2] outdoor people and vehicle tracking (two synchronised views; includes omnidirectional and moving camera) (annotation available), available here.  Details on the dataset specification is available here.

• [3] indoor people tracking (and counting) and hand posture classification data (annotated), available here.  Details on the dataset specification is available here.

• [4] annotation of a smart meeting (annotation available).  Includes facial expressions, gaze and gesture/action, available here.  Details on the dataset specification is available here.

• [5] outdoor people tracking - football data (two views), available here.  Details on the dataset specification is available here.

The results of processing the datasets should be submitted in XML format.  The details (available soon) will be given here. 

Papers are solicted which

• (1) Describe an approach to target detection and/or tracking and reports results based on the datasets provided for this workshop available here OR based on previously published PETS datasets (see above), The paper should include evaluation of the results obtained on the data.

• (2) Contribute to performance evaluation methodologies for tracking and/or surveillance.  This can be a general paper outlining, for example, a new methodology for evaluation, or performance evaluation best practice etc.  It is not necessary to explictly consider the datasets provided for the workshop, however you are encouraged to use previous PETS datasets (see above) available here.

All printed material, including illustrations, must be kept within a print area of 6-7/8 inches (17.5 cm) wide by 8-7/8 inches (22.5 cm) high. A detailed description of the format is given in the IEEE Author Guidelines. There is a simple LaTex template for this format that may assist you in meeting this format. Or see the MS word template. The submitted file should be in Postscript or PDF format.

Please submit your paper as an email attachment to pets2005@visualsurveillance.org
The paper should be named by concatenating the names of the authors' last names followed by a single digit in the case of multiple submissions by the same authors.

Do not identify authors in the submitted paper.  All reviewing will be double blind, so the paper must not include any information which allows the authors to be identified.  The names of the authors, however, should be entered along with other details in the email submission. 

For guidance on style, you may like to consult the electronic proceedings for PETS'2000 located here. The electronic proceedings for PETS'2001 are located here.

Dr James Ferryman

Computational Vision Group
Department of Computer Science
The University of Reading
Whiteknights 
PO Box 225 Reading Berkshire 
England 

Phone: +44 118 931 6697 
Fax: +44 118 975 1994 
E-mail: J.M.Ferryman@reading.ac.uk 

Last modified: at Nov 13 19:56:58 GMT 2004 
pets2005@visualsurveillance.org
pets2005.visualsurveillance.org