Video genre verification using both acoustic and visual modes
Matthew Roach, J. Mason, Li-Qun Xu · 2004
This paper reports on the verification of the video genre: sport, cartoon, news, commercial and music. Results for the two modes, acoustic and visual, and for combined modes show an average equal error rate (ERR) of 16%, 15% and 10%, respectively. These reflect verification accuracy and as such are believed to be the first of their kind; previously published work has focused on closed set identification, assuming the video is known to belong to one of a fixed set. The results also demonstrate the influence of the genre to be classified: the best performance for the visual mode has an EER of 4% (cartoons), and the best performance for the acoustic mode has EER of 0.6% (news). Finally, the combination of the modes presents a more consistent accuracy across the five genre with an EER of 10%.