By Martha Larson
MediaEval is an international multimedia benchmarking initiative that offers tasks to the multimedia community that are related to the human and social aspects of multimedia. The focus is on addressing new challenges in the area of multimedia search and indexing that allow researchers to make full use of multimedia techniques that simultaneously exploit multiple modalities.
A series of interesting tasks is currently underway in MediaEval 2013. As every year, the selection of tasks is made using a community-wide survey that gauges what multimedia researchers would find most interesting and useful. A new task to watch closely this year is Search and Hyperlinking of Television Content, which follows on the heels of a very successful pilot last year. The other main tasks running this year are:
- Social Event Detection for Social Multimedia
- Placing: Geo-coordinate Prediction for Social Multimedia
- Violent Scenes Detection in Film (Affect Task)
- Visual Privacy: Preserving Privacy in Surveillance Videos
- Spoken Web Search: Spoken Term Detection for Low Resource Languages
The tagline of the MediaEval Multimedia Benchmark is: “The ‘multi’ in multimedia: speech, audio, visual content, tags, users, context”. This tagline explains the inspiration behind the choice of the Brave New Tasks, which are running for the first time this year. Here, we would like to highlight Question Answering for the Spoken Web, which builds on the Spoken Web Search tasks mentioned above. This task is a joint effort between MediaEval and the Forum for Information Retrieval Evaluation, a India-based information retrieval benchmark. MediaEval believes strongly in collaboration and complementarity between benchmarks and we hope that this task will help us to better understand how joint-tasks should be best designed and coordinated.
The other Brave New Tasks at MediaEval this year are:
- Soundtrack Selection for Commercials a task from MusiClef
- Similar Segments of Social Speech
- Retrieving Diverse Social Images
- Emotion in Music
- Crowdsourcing for Social Multimedia
The MediaEval 2013 season culminates with the MediaEval 2013 worksop, which will take place in Barcelona, Catalunya, Spain, on Friday-Saturday 18-19 October 2013. Note that this is just before ACM Multimedia 201, which will be held Monday-Friday 21-25 October 2013, also in Barcelona. We are currently working on finalizing the registration site for the workshop and it will open very soon and will be announced on the MediaEval website.
In order to further foster our understanding and appreciation of user-generated multimedia, each year we designate a MediaEval filmmaker to make a YouTube video about the workshop. The MediaEval 2012 workshop video was made by John N.A. Brown and has recently appeared online.
John’s decided to focus on the sense of community and the laughter than he observed at the workshop. Interestingly, his focus recalls the work done at MediaEval 2010 on the role of laughter in social video, see:
We hope that this video inspires your to join us in Barcelona.