Successfully reported this slideshow.
We use your LinkedIn profile and activity data to personalize ads and to show you more relevant ads. You can change your ad preferences anytime.
Maria Eskevich, Quoc-Minh Bui, Hoang-An Le,
Benoit Huet
Multimedia Department
EURECOM
Sophia-Antipolis, France
Exploring V...
Motivation
 CURRENTLY: Constantly growing quantity of
multimedia content produced by both
professionals and individual us...
Task and Challenges
 We envisage the users to be interested in creating their
own path within the multimedia collection b...
Insights in Hyperlinking
 Hyperlinking
Creating “links” between media
 Video Hyperlinking
video to video
video fragme...
Related experiments: Search and
Hyperlinking @ MediaEval/TRECVid
 Different techniques based on:
 Video segmentation met...
Search and Hyperlinking @
MediaEval/TRECVid
 Our solution:
 Scene segmentation technique based on visual
and temporal co...
System overview
Broadcast Media: BBC content
Manually transcribed
subtitles
Metadata:
title, cast, description, broadcast ...
151 Visual Concepts (TrecVid 2012)
 3_Or_More_People
 Actor
 Adult
 Adult_Female_Human
 Adult_Male_Human
 Airplane
...
Scene segmentation
 Scene : group of shots based on content similarity
and temporal consistency among shots;
 Content si...
Handling Visual Concepts in Solr
10/30/15 SLAM Workshop @ ACM MM 2015, Brisbane, Australia 10
http://localhost:8983/solr/c...
Results (S&H 2014)
10/30/15 SLAM Workshop @ ACM MM 2015, Brisbane, Australia 11
Results discussion
 Text (Audio) performs best on the 2014
MediaEval Hyperlinking task
 Using visual features isn’t stra...
DEMO
 Hyper Video Browser
10/30/15 SLAM Workshop @ ACM MM 2015, Brisbane, Australia 13
Search/Browse Hyperlinking
Conclusion and Future work
 We proposed and evaluated an approach to include
visual properties in the search of video seg...
References
 E. Apostolidis, V. Mezaris, M. Sahuguet, B. Huet, B. Cervenkova, D. Stein, S. Eickeler, J.
L. Redondo Garcia,...
Questions?
Thank you for your
attention!
Upcoming SlideShare
Loading in …5
×

Exploring Video Hyperlinking in Broadcast Media

1,224 views

Published on

Multimedia content produced on a daily basis and in constantly growing quantity by professionals and individual users,
requires creation of navigation systems that allow access to
this data on different levels of granularity in order to con-
tribute to further discovery of a topic of interest for the user
or to facilitate individual user browsing within a collection.
In this paper we describe our approach to enable users
to browse through the multimedia collection. We implement the hyperlinking approach that uses the fine-grained
segmentation of the visual content based on the scene segmentation, as well as available metadata, transcripts, and
information about extracted visual concepts.
The approach was tested at the MediaEval Search and
Hyperlinking 2014 evaluation task, where it has shown its effectiveness at locating accurately relevant content in a large
media archive.

Published in: Technology
  • Be the first to comment

Exploring Video Hyperlinking in Broadcast Media

  1. 1. Maria Eskevich, Quoc-Minh Bui, Hoang-An Le, Benoit Huet Multimedia Department EURECOM Sophia-Antipolis, France Exploring Video Hyperlinking in Broadcast Media
  2. 2. Motivation  CURRENTLY: Constantly growing quantity of multimedia content produced by both professionals and individual users.  NEED: navigation systems that allow access to this data on different levels of granularity in order to contribute to further discovery of a topic of interest for the user or to facilitate individual user browsing within a collection. 10/30/15 SLAM Workshop @ ACM MM 2015, Brisbane, Australia 2
  3. 3. Task and Challenges  We envisage the users to be interested in creating their own path within the multimedia collection based on their level of awareness/knowledge of it, and tasks.  How do we create a hyperlinked collection to allow each individual user to follow their own path of interest within?  Why is it challenging?  Lack of interpretability of the content: Overall textual description on the video level which leads to: Retrieval process is based on text features, while the rich multimodality of videos is not exploited; Not possible to partially retrieve a video, a specific fragment without having to watch the entire video.  Archives are not static: Need for dynamic creation of links 10/30/15 SLAM Workshop @ ACM MM 2015, Brisbane, Australia 3
  4. 4. Insights in Hyperlinking  Hyperlinking Creating “links” between media  Video Hyperlinking video to video video fragment to video fragment 10/30/15 SLAM Workshop @ ACM MM 2015, Brisbane, Australia 4
  5. 5. Related experiments: Search and Hyperlinking @ MediaEval/TRECVid  Different techniques based on:  Video segmentation method:  Fixed-length units (same length, sentences, speech segments) with further adjustment to match full sentences, using speech segment boundaries and pauses. they build a probabilistic framework to model the importance of words and refine segment boundaries accordingly;  Meaningful segments, based on topics derived from transcripts: classification trees to define the starting and ending times of these segments; lexical cohesion within segment.  Features used for retrieval:  Text similarity (vector-based models and TF-IDF weightings);  Named entities or synonyms;  Visual information: visual concepts or SURF and SIFT features, detected at the shot level. 10/30/15 SLAM Workshop @ ACM MM 2015, Brisbane, Australia 5
  6. 6. Search and Hyperlinking @ MediaEval/TRECVid  Our solution:  Scene segmentation technique based on visual and temporal coherence of the video segments that constitute the scenes of the video.  Experiment with hybrid segmentation Visual, Topic and Temporal Coherence  Use visual features to improve the ranking of results of the hyperlinking including visual analysis during the search process. 10/30/15 SLAM Workshop @ ACM MM 2015, Brisbane, Australia 6
  7. 7. System overview Broadcast Media: BBC content Manually transcribed subtitles Metadata: title, cast, description, broadcast time Shot segmentation, keyframes Visual content analysis: Scene segmentation Concept detection on shot level Lucene/Solr: Indexing/Retrieval on shot level Media database: 76 214 fragments Webservice: (HTML5/AJAX/PHP) Videos: 2323 items/1697 hours User Interface Shot segmentation, keyframe extraction Result list: Media fragments level
  8. 8. 151 Visual Concepts (TrecVid 2012)  3_Or_More_People  Actor  Adult  Adult_Female_Human  Adult_Male_Human  Airplane  Airplane_Flying  Airport_Or_Airfield  Anchorperson  Animal  Animation_Cartoon  Armed_Person  Athlete  Baby  Baseball  Basketball  Beach  Bicycles  Bicycling  Birds  Boat_Ship  Boy • Building • Bus • Car • Car_Racing • Cats • Cattle • Chair • Charts • Child • Church • City • Cityscape • Classroom • Clouds • Construction_Vehicles • Court • Crowd • Dancing • Daytime_Outdoor • Demonstration_Or_Protest • Desert • Dogs • Emergency_Vehicles • Explosion_Fire • Face • Factory • Female-Human-Face-Closeup • Female_Anchor • Female_Human_Face • Female_Person • Female_Reporter • Fields • Flags • Flowers • Football • Forest • Girl • Golf • Graphic • Greeting • Ground_Combat • Gun • Handshaking • Harbors • Helicopter_Hovering • Helicopters • Highway • Hill • Hockey • Horse • Hospital • Human_Young_Adult • Indoor • Insect • Kitchen • Laboratory • Landscape • Machine_Guns • Male-Human-Face-Closeup • Male_Anchor • Male_Human_Face • Male_Person • Male_Reporter • Man_Wearing_A_Suit • Maps • Meeting • … 10/30/15 SLAM Workshop @ ACM MM 2015, Brisbane, Australia 8
  9. 9. Scene segmentation  Scene : group of shots based on content similarity and temporal consistency among shots;  Content similarity : visual similarity between HSV histograms extracted from the keyframes of different shots;  Grouping is performed using two extensions of the Scene Transition Graph (STG):  reduces the computational cost of STG-based shot grouping by considering shot linking transitivity,  builds on the former to construct a probabilistic framework that alleviates the need for manual STG parameter selection. [Apostolidis et al. “Automatic fine-grained hyperlinking of videos within a closed collection using scene segmentation.” ACM MM 2014] 10/30/15 SLAM Workshop @ ACM MM 2015, Brisbane, Australia 9
  10. 10. Handling Visual Concepts in Solr 10/30/15 SLAM Workshop @ ACM MM 2015, Brisbane, Australia 10 http://localhost:8983/solr/collection_mediaEval/select?q=text:(Children out on poetry trip Exploration of poetry by school children Poem writing) Animal:[0.2 TO 1] Building:[0.2 TO 1]  Schema = structure of document using fields of different types  Query <doc> <field name="id"> 20080401_013000_bbcfour_legends_marty_feldman_six_degrees_of#t=399,402</field> <field name="begin">00:06:39.644</field> <field name="end">00:06:42.285</field> <field name="videoId">20080401_013000_bbcfour_legends_marty_feldman_six_degrees_of</field> <field name="subtitle">'It was very, very successful.'</field> <field name="Actor">0.143</field> <field name="Adult">0.239</field> <field name="Animal">0.0572</field> </doc>
  11. 11. Results (S&H 2014) 10/30/15 SLAM Workshop @ ACM MM 2015, Brisbane, Australia 11
  12. 12. Results discussion  Text (Audio) performs best on the 2014 MediaEval Hyperlinking task  Using visual features isn’t straightforward  The ‘MoreLikeThis’ approach is outperformed by the Text only search Possibly due to query formulation differences Sentences vs Keywords  Visual Scenes outperform other fragmentation levels (Topic, Sentences) 10/30/15 SLAM Workshop @ ACM MM 2015, Brisbane, Australia 12 [Safadi et al. “When textual and visual information join forces for multimedia retrieval”, ICMR 2014]
  13. 13. DEMO  Hyper Video Browser 10/30/15 SLAM Workshop @ ACM MM 2015, Brisbane, Australia 13 Search/Browse Hyperlinking
  14. 14. Conclusion and Future work  We proposed and evaluated an approach to include visual properties in the search of video segments for hyperlinking.  Experimental results show that :  Visual properties provide meaningful cues for segmenting the video  Mapping text-based queries to visual concepts is not easy  Automatic selection of relevant concepts is required (human intervention is impractical and does not necessarily lead to perfect results)  Operating at the keyword level rather than full text offers improvements  Current/Future work: incorporate query semantics when identifying key visual semantic concepts  based on named entity recognition approaches and keyword/visual concepts co-occurrences 10/30/15 SLAM Workshop @ ACM MM 2015, Brisbane, Australia 14
  15. 15. References  E. Apostolidis, V. Mezaris, M. Sahuguet, B. Huet, B. Cervenkova, D. Stein, S. Eickeler, J. L. Redondo Garcia,R. Troncy, and L. Pikora. “Automatic ne-grained hyperlinking of videos within a closed collection using scene segmentation”. In ACMMM 2014, 22nd ACM International Conference on Multimedia, Orlando, Florida, USA, 11 2014.  H. Le, Q. Bui, B. Huet, B. Cervenkova, J. Bouchner, E. Apostolidis, F. Markatopoulou, A. Pournaras, V. Mezaris, D. Stein, S. Eickeler, and M. Stadtschnitzer,.“LinkedTV at MediaEval 2014 Search and Hyperlinking Task”. In Proceedings of MediaEval 2014 Workshop, 2014.  R. J. F. Ordelman, M. Eskevich, R. Aly, B. Huet, and G. J. F. Jones. “Dening and evaluating video hyperlinking for navigating multimedia archives”. In Proceedings of the 24th International Conference on World Wide Web Companion, WWW 2015, Florence, Italy, May 18-22, 2015 - Companion Volume, pages 727-732, 2015.  M. Sahuguet, B. Huet, B. Cervenkova, E. Apostolidis, V. Mezaris, D. Stein, S. Eickeler, J. L. Redondo Garcia, and L. Pikora. “LinkedTV at MediaEval 2013 search and hyperlinking task”. In MediaEval 2013 Workshop, Barcelona, Spain, 10 2013.  B. Safadi, M. Sahuguet, and B. Huet. “When textual and visual information join forces for multimedia retrieval”. In Proceedings of International Conference on Multimedia Retrieval (ICMR '14). ACM, New York, NY, USA, , Pages 265 , 8 pages, 2014. 10/30/15 SLAM Workshop @ ACM MM 2015, Brisbane, Australia 15
  16. 16. Questions? Thank you for your attention!

×