Multimedia Knowledge and Social Media Analytics Laboratory

CERTH is the best in detecting whether you are sitting down or walking!

CERTH participated to the Semantic Indexing (SIN) Task of TRECVID 2010 with its video concept detectors. CERTH’s concept detectors produced the best results, among the submissions made by all institutions participating to TRECVID (101 runs from 30 different institutions), for two of the concepts evaluated this year ("walking", "sitting down"). The latest technical details on the approach that was developed and used for building these detectors are reported in the following publication: V. Mezaris, A. Dimou, I. Kompatsiaris, "On the use of feature tracks for dynamic concept detection in video", Proc. IEEE International Conference on Image Processing (ICIP 2010), Hong Kong, China, September 2010.