TRECVid Surveillance Event Detection Evaluation Track

The Surveillance Event Detection (SED) evaluation track is part of the TRECVid Evaluation and is intended to help promote technology development for event detection in video surveillance. The goal is to leverage machine learning techniques for detecting a pre-defined set of events. The evaluation tasks are focused on detecting events in airport surveillance video.

The SED evaluation track has been a part of the TRECVid evaluation series since 2008. There have been nine open evaluations, in 2008 and 2009, 2010, 2011, 2012, 2013, 2014, 2015 and 2016. The 2017 evaluation is currently underway for the 2017 TRECVid evaluation cycle.

Information about the data used in the SED evaluations, see: Licensing/Participation Agreements

Results of past SED evaluations can be found on the TRECVid Publications page. In addition, a more detailed analysis of the 2008 SED results can be found in the summary paper "The TRECVid 2008 Event Detection Evaluation" presented at the 2009 workshop on Applications of Computer Vision (WACV).  

Yearly evaluation pages

[2008] [2009] [2010] [2011] [2012] [2014] [2015] [2016] [2017]