British Machine Vision Association and Society for Pattern Recognition

Call for Participation: Deadline 30'th October

Security and Surveillance: performance evaluation

One Day BMVA symposium at the British Computer Society, Southampton
Street, London, UK on 12th December 2007

Chair: Andrea Cavallaro - Queen Mary, University of London

Performance evaluation and validation is an important open problem for
computer vision algorithms applied to security and surveillance
applications due to the lack of commonly accessible datasets and
commonly accepted evaluation protocols. Although in recent years
important efforts have been devoted to the design of evaluation metrics
and the comparison of algorithms (e.g., PETS, ETISEO, CLEAR, AMI,
CAVIAR), these are not yet taken up by the research community at large.
Moreover, computer vision algorithms need be evaluated with large test
corpora containing significant statistical data variability as the
accuracy of their results is highly data dependent.

The aim of this meeting is to provide a forum for the discussion of
recent algorithms, results, protocols and datasets for the evaluation
and validation of computer vision algorithms for security and
surveillance. Contributions describing recent work on performance
evaluation of object detection, object classification, target tracking,
activity analysis and event recognition are welcome. Manuscripts
describing new or consolidated datasets and metrics for the evaluation
and comparison of algorithms are also sought. Potential topics include,
but are not limited to: metrics for the evaluation computer vision
algorithms applied to surveillance; benchmarking protocols and their
application; metrics for evaluating the complexity of a scene/situation;
metrics for evaluating the performance of single and multi-target
trackers; dataset distribution and privacy issues; ground truth
definition and XML schemas; experience from previous evaluation
campaigns.

Please submit an extended summary of about one A4-sized page (no longer
than two pages) in length (PDF preferred). Send contributions by email
attachment (1Mb max please!) to Andrea Cavallaro
( mailto:andrea.cavallaro@elec.qmul.ac.uk ) by 30th October 2007.

Full Meeting Schedule: http://www.bmva.ac.uk/meetings

Simon Prince, Department of Computer Science
University College London, Gower Street, London, WC1E 6BT
Tel. 020 7679 3692, Fax. 020 7387 1397
http://www.cs.ucl.ac.uk/staff/S.Prince/