Petra Galuščáková presents Multimodal Features for Search and Hyperlinking of TV Content
On 2015-02-12 11:00
at G205, Karlovo náměstí 13, Praha 2
In the talk, I will discuss content-based retrieval in audio-visual collections.
I will focus on retrieval of relevant segments of videos using a textual query.
In addition, I will describe techniques for an automatic detection of hyperlinks
between similar video segments within audio-visual collections. In the
experiments, text-based similarity, which uses subtitles and automatic
transcripts, is combined with similarity based on the visual and prosodic
features. The employment of the visual similarity achieves a constant
improvement, while the employment of the prosodic similarity shows a small but
promising improvement. Our retrieval system ranked first in the MediaEval 2014
Search and Hyperlinking shared task. The experiments are performed on almost
4000 hours of BBC broadcast video.
I will focus on retrieval of relevant segments of videos using a textual query.
In addition, I will describe techniques for an automatic detection of hyperlinks
between similar video segments within audio-visual collections. In the
experiments, text-based similarity, which uses subtitles and automatic
transcripts, is combined with similarity based on the visual and prosodic
features. The employment of the visual similarity achieves a constant
improvement, while the employment of the prosodic similarity shows a small but
promising improvement. Our retrieval system ranked first in the MediaEval 2014
Search and Hyperlinking shared task. The experiments are performed on almost
4000 hours of BBC broadcast video.