[ Skip to the content ]

Institute of Formal and Applied Linguistics

at Faculty of Mathematics and Physics, Charles University, Prague, Czech Republic


[ Back to the navigation ]

Publication


Year 2016
Type in proceedings without ISBN
Status published
Language English
Author(s) Galuščáková, Petra Batko, Michal Kruliš, Martin Lokoč, Jakub Novák, David Pecina, Pavel
Title CUNI at TRECVID 2015 Video Hyperlinking Task
Czech title CUNI na benchmarku TRECVID 2015 Video Hyperlinking Task
Proceedings 2016: Gaithersburg, MD, USA: TRECVID: 2015 TREC Video Retrieval Evaluation Notebook Papers and Slides
Number 2015
Pages range 1-7
How published online
URL http://www-nlpir.nist.gov/projects/tvpubs/tv.pubs.15.org.html
Supported by 2012-2018 GBP103/12/G084 (Centrum pro multi-modální interpretaci dat velkého rozsahu) 2013-2014 GAUK 9209/2013 (Interaktívne spôsoby vyhľadávania v audiovizuálnych dialógových korpusoch) 2015 SVV 260 224 (Teoretické základy informatiky a výpočetní lingvistiky) 2012-2016 PRVOUK P46 (Informatika) 2016 SVV 260 333 (Teoretické základy informatiky a výpočetní lingvistiky)
Czech abstract V článku popisujeme náš postup aplikovaný v benchmarku TRECVID 2015 v úlohe Video Hyperlinking. Náš prístup kombinuje podobnosť textov vypočítanú z titulkov, vizuálnu podobnosť medzi zábermi vypočítanú pomocou Feature Signatures a informáciou o tom, či query segment a vyhľadaný segment pochádzajú z toho istého televízneho programu. Všetky experimenty boli odladené a otestované na kolekcii 2500 hodín televíznych programov poskytnutých BBC.
English abstract In this paper, we present our approach used in the TRECVID 2015 Video Hyperlinking Task. Our approach combines text-based similarity calculated on subtitles, visual similarity between keyframes calculated using Feature Signatures, and preference whether the query and retrieved answer come from the same TV series. All experiments were tuned and tested on about 2500 hours of BBC TV programmes. Our Baseline run exploits fixed-length segmentation, text-based retrieval of subtitles, and query expansion which utilizes metadata, context, in-formation about music and artist contained in the query segment and visual concepts. The Series run combines the Baseline run with weighting based on information whether the query and data segment come from the same TV series. The FS run combines the Baseline run with the similarity between query and data keyframes calculated using Feature Signatures. The FSSeriesRerank run is based on the FS run on which we applied reranking which, again, uses information about the TV series. The Series run significantly outperforms the FSSeriesRerank run. Both these runs are significantly inferior to our Baseline run in terms of all our reported measures. The FS run outperforms the Baseline run in terms of all measures but it is significantly better than the Baseline run only in terms of the MAP score. Our test results confirm that employment of visual similarity can improve video retrieval based on information contained in subtitles but information about TV series which was most helpful in our training experiments did not lead to further improvements.
Specialization linguistics ("jazykověda")
Confidentiality default – not confidential
Open access no
Address* Gaithersburg, MD, USA
Publisher* NIST
Institution* NIST
Creator: Common Account
Created: 7/4/16 5:47 PM
Modifier: Almighty Admin
Modified: 2/25/17 10:07 PM
***

Content, Design & Functionality: ÚFAL, 2006–2016. Page generated: Sat Nov 18 03:39:14 CET 2017

[ Back to the navigation ] [ Back to the content ]

100% OpenAIRE compliant