MediaEval 2012 Working Notes Proceedings
The Working Notes Proceedings of the MediaEval 2012 Workshop are now available online at:

Example of how to cite a paper from the proceedings:

Joder, C., Weninger, F., Wollmer, M. and Schuller, B. The TUM Cumulative DTW Approach for the Mediaeval 2012 Spoken Web Search Task. Working Notes Proceedings of the MediaEval 2012 Workshop, Pisa, Italy, October 4-5, 2012,, ISSN 1613-0073,

MediaEval 2012 Workshop

Slides from the workshop are available at:

Photos from the workshop are at:

Program of the MediaEval 2012 Workshop
4-5 October 2012
Santa Croce in Fossabanda Pisa, Italy

Thursday 4 October 2012
8:30-9:00 Registration

9:00 Workshop opening (Brief words of welcome from Martha Larson and Gareth Jones) slides

9:00-10:30 Placing Task
Session Chair: Pascal Kelm (TU Berlin, Germany)
0. Placing Task overview: "Working Notes for the Placing Task at MediaEval 2012" (presenter: Adam Rae. Yahoo! Research, Spain) slides paper
1. "The 2012 ICSI/Berkeley Video Location Estimation System" (presenter: Jaeyoung Choi, ICSI, USA) slides paper
2. "Ghent and Cardiff University at the 2012 Placing Task" (presenter: Olivier van Laere, Ghent University, Belgium) slides paper
3. "Preliminary exploration of the use of geographical information for content-based geo-tagging of social video" (presenter: Xinchao Li, TU Delft, Netherlands) slides paper
4. "How INRIA identifies Geographic Location of a Video" (presenter: Michele Trevisiol, Universitat Pompeu Fabra, Spain) slides paper
5. "How Spatial Segmentation improves the Multimodal Geo-Tagging" (presenter: Pascal Kelm, TU Berlin, Germany) slides paper
Also see University of Campinas, Brazil, slides paper and CEA LIST, France paper

10:30-11:00 Coffee

11:00-11:15 MediaEval 2012: The fifth year of our multimedia benchmark initiative (Martha Larson)

11:15 - 12:30 Social Event Detection Task
Session Chair: Vasileios Mezaris (CERTH-ITI, Greece)
0. Social Event Detection Task overview: "SED at MediaEval 2012: Challenges, Dataset and Evaluation" (presenter: Raphael Troncy, EURECOM, France) slides paper
1. "Event Detection via LDA for the MediaEval2012 SED Task" (presenter: Konstantinos Vavliakis, Aristotle University of Thessaloniki and CERTH-ITI) slides paper
2. "CERTH @ MediaEval 2012 Social Event Detection Task" (presenter: Manos Schinas, CERTH-ITI, Greece) slides paper
3. "The Watershed-based Social Events Detection Method with Support from External Data Sources" (presenter: Minh Son Dao, University of Trento, Italy) slides paper
4. "QMUL @ MediaEval 2012: Social Event Detection in Collaborative Photo Collections" (presenter: Markus Brenner, Queen Mary University of London, UK) slides paper
See also Vienna University of Technology/University of Vienna, Austria paper

12:30 - 13:30 Lunch

13:30 - 13:45 Invited Talk by Jana Eggink, BBC Research and Development

13:45-15:30 Affect Task: Violent Scenes Detection Task
Session Chair: Mohammad Soleymani (Imperial College London, UK)
0. Affect Task: Violent Scenes Detection Task overview: "The MediaEval 2012 Affect Task: Violent Scenes Detection" (presenter: Claire-Helene Demarty, Technicolor, France) slides paper
1. "LIG at MediaEval 2012 Affect Task: Use of a Generic Method" (presenter: Nadia Derbas, Laboratoire d'Informatique de Grenoble, France) slides paper
2. "ARF @ MediaEval 2012: An Uninformed Approach to Violence Detection in Hollywood Movies" (presenter: Bogdan Ionescu, University Politehnica of Bucharest, Romania) slides paper
3. "DAI Lab at MediaEval 2012 Affect Task: The Detection of Violent Scenes using Affective Features" (presenter: Esra Acar, TU Berlin, Germany) slides paper
4. "The Shanghai-Hongkong Team at MediaEval2012: Violent Scene Detection Using Trajectory-based Features" (presenter: Tan Chun Chet, City University of Hong Kong) slides paper
5. "Violent Scenes Detection with Large, Brute-forced Acoustic and Visual Feature Sets" (presenter: Felix Weninger, TU Munich, Germany) slides paper
6. "Technicolor/INRIA/Imperial College London at the MediaEval 2012 Violent Scene Detection Task" (presenter: Cédric Penet, Technicolor, France) slides paper
Also see NII, Japan slides paper and Universite de Toulon, Aix Marseile Universite, Institute Universitaire de France slides paper

15:30-16:00 Coffee

16:00-18:15 Spoken Web Search Task
Session Chair: Guillaume Gravier (CNRS/IRISA, France)
0. Spoken Web Search Task overview "The Spoken Web Search Task" (presenter: Florian Metze, CMU, USA) slides paper
1. "The L2F Spoken Web Search system for Mediaeval 2012" (presenter: Alberto Abad, INESC-ID, Portugal) slides paper
2. "BUT2012 Approaches for Spoken Web Search---MediaEval 2012" (presenter: Igor Szoke, Brno University of Technology, Czech Republic) slides paper
3. "Telefonica Research System for the Spoken Web Search task at Mediaeval 2012" (presenter: Xavier Anguera, Telefonica Research, Spain) slides paper
4. "The TUM Cumulative DTW Approach for the Mediaeval 2012 Spoken Web Search Task" (presenter: Felix Weninger, TU Munich, Germany) slides paper
5. "GTTS System for the Spoken Web Search Task at MediaEval 2012" (presenter: Luis Javier Rodríguez-Fuentes, University of the Basque Country, Spain) slides paper
6. "TUKE MediaEval 2012: Spoken Web Search using DTW and Unsupervised SVM" (presenter: Jozef Vavrek, TU Košice, Slovakia) slides paper
7. "CUHK System for the Spoken Web Search task at Mediaeval 2012" (presenter: Haipeng Wang, Chinese University of Hong Kong) slides paper
8. "ARF @ MediaEval 2012: A Romanian ASR-based Approach to Spoken Term Detection" (presenter: Andi Buzo, University Politehnica of Bucharest, Romania) slides paper
Also see Johns Hopkins University, USA slides paper

18:30-20:00 Poster session and aperativo
(Session Chair: Xavi Anguera, Telefonica Research, Spain)

20:00 Workshop dinner (i.e., the "social event")

Friday 5 October 2012
8:00-9:00 (Optional) Early Morning Lightening: Optional session for those interested in proposing tasks for next year
Session Chair: Gareth Jones (Dublin City University, Ireland)

9:00 Opening of Day 2

9:15-9:30 Invited Talk by Nicola Ferro, University of Padova (PROMISE Co-ordinator) slides

9:30-10:30 Visual Privacy Task

Session Chair: Marco Tiemann, University of Reading, UK
0. Visual Privacy Task overview "Overview of MediaEval 2012 Visual Privacy Task " (presenter: Tomas Piatrik, Queen Mary University of London, UK) slides paper paper
1. "MediaEval 2012 Visual Privacy Task: Applying Transform-domain Scrambling to Automatically Detected Faces" (presenter: Pavel Kurshunov, EPFL, Switzerland) slides paper
2. "MediaEval 2012: Scrambling Faces for Flexible Privacy Preservation using Image Background Self-similarities" (presenter: Claudia Araimo, EURECOM, France) slides paper
3. "MediaEval 2012 Visual Privacy Task: Privacy and Intelligibility through Pixellation and Edge Detection" (presenter: Mathieu Louis Eining, University of Reading, UK) slides paper

10:30-11:00 Coffee

11:00-11:15 MediaEval Code of Conduct
, Invited Talk, by Adam Rae, Yahoo! Research, Spain

11:15-12:30 Tagging Task
Session Chair: Christoph Kofler, TU Delft, Netherlands
0. Tagging Task overview "Overview of the MediaEval 2012 Tagging Task" (presenter: Isabelle Ferrané, University of Toulouse, France) slides paper
1. "ARF @ MediaEval 2012: Multimodal Video Classification" (presenter: Bogdan Ionescu, University Politehnica of Bucharest, Romania) slides paper
2. "KIT at MediaEval 2012 – Content–based Genre Classification with Visual Cues" (presenter: Tomas Semela, Karlsruhe Institute of Technology, Germany) slides paper
3. "TUB @ MediaEval 2012 Tagging Task: Feature Selection Methods for Bag-of-(visual)-Words Approaches" (presenter: Sebastian Schmiedeke, TU Berlin, Germany) slides paper
4. "TUD at MediaEval 2012 genre tagging task" (presenter: Martha Larson, TU Delft, Netherlands) slides paper1 paper2
See also University of Campinas and University of Minas Gerais, Brazil slides paper

12:30-13:30 Lunch

13:30-14:00 Brave New Task: User Account Matching
Session Chair: Roeland Ordelman (Netherlands Institute for Sound and Vision and Twente University, Netherlands)
1. Presentation by Claudia Hauff, TU Delft, Netherlands slides paper
2. Presentation by Jaeyong Choi, ICSI, USA

14:00-15:30 Brave New Task: Search and Hyperlinking
Session Chair: Robin Aly (Twente University, Netherlands)
0. Search and Hyperlinking Task overview: "Search and Hyperlinking Task at MediaEval 2012" (presenter: Maria Eskevich, Dublin City University, Ireland) slides
1. "CUNI at MediaEval 2012 Search and Hyperlinking Task" (presenter: Petra Galuscakova, Charles University, Slovakia) slides paper
2. "Ghent University-IBBT at MediaEval 2012 Search and Hyperlinking: Semantic Similarity using Named Entities" (presenter: Tom De Nies, Ghent University, Belgium) slides paper
3. "UTwente does Brave New Tasks for MediaEval 2012: Searching and Linking" (presenter: Danish Nadeem, Twente University, Netherlands) paper
4. "IRISA at MediaEval 2012: Search and Hyperlinking Task" (presenter: Camille Guinaudeau, IRISA, France) paper
5. "DCU at MediaEval 2012 Search and Hyperlinking" (presenter: Shu Chen, Dubin City University, Ireland) slides paper1 paper2

15:30-16:00 Coffee

16:00-16:15 The Virtual Kitchen for benchmarking: Invited Talk, Florian Metze, CMU, USA slides

16:15-16:45 Brave New Task: Musiclef Multimodal Music Tagging (presenter: Cynthia Liem, TU Delft, Netherlands) slides paper
Also see LUTIN UserLab, France paper

16:45-17:00 Workshop closing (Martha Larson and Mohammad Soleymani) slides

MediaEval 2012 Workshop Organizing Committee

Proceedings Chairs: Sebastian Schmiedeke and Pascal Kelm, Technische Universität Berlin, Germany
Practitioners Chair: Xavier Anguera, Telefonica, Spain
Awards Chair: Mohammad Soleymani, Imperial College London, UK
Multimedia Chair: Florian Metze, CMU, USA (Supported by Babak Loni, TU Delft, Netherlands)
Logistics Chair: Saskia Peters, TU Delft, Netherlands
Local Organization support: Alessio Bazzica, Università degli Studi di Firenze, Italy
General Chair and contact person: Martha Larson, TU Delft, Netherlands

Overview of MediaEval 2012 Tasks

Placing Task
Participants place Flickr videos on a map,' i.e., assign geo-coordinates to videos using one or more of: Flickr metadata, visual content, audio content, social information. Read more...

Social Event Detection Task
This task requires participants to discover events and detect media items that are related to either a specific social event or an event-class of interest. Read more...

Spoken Web Search
This task involves searching FOR audio content WITHIN audio content USING an audio content query. This task is particularly interesting for speech researchers in the area of spoken term detection. Read more...

Tagging Task
This task requires participants to automatically assign tags to Internet videos using features derived from speech, audio, visual content or associated textual or social information. This year we will again focus on tags that reflect the genre of the video. Read more...

Affect Task: Violence Detection Task
This task requires participants to deploy multimodal features to automatically detect of portions of movies containing violent material. Read more...

Visual Privacy
Participants blur human faces in video with the goal of obscuring privacy while simultaneously assuring that the resulting visual impression is maximally acceptable to human viewers. Read more...

Brave New Tasks
Additionally, User Account Matching and Search and Hyperlinking will run as Brave New Tasks.

2012 schedule
The exact dates of the data releases are set by the task organizers and are listed on the description pages of the individual tasks.

mid-February: MediaEval 2012 tasks announced
1 March: MediaEval 2012 sign up opens
31 May: Last day for regular sign up (late sign-up will stay open a bit longer)
May-June: Development data release
June-July: Test data release
August-September: Run submission deadline
mid-September: Release of results
End September: Two-page working notes papers due
4-5 October: MediaEval 2012 Workshop in Pisa, Italy
The workshop is timed so that people can attend the 12th European Conference on Computer Vision ECCV 2012 (7-13 October) in the same trip.

For questions please contact Martha Larson m.a.larson (at)