MediaEval (Benchmarking Initiative for Multimedia Evaluation) offers shared-tasks to the multimedia research community involving images, text, video, and audio/music. The tasks address cutting-edge multimedia challenges (multimedia retrieval, access and exploration) with a clear human or social aspect.
Further information and the link for registration is available at: http://www.multimediaeval.org/mediaeval2017
Short descriptions of the tasks are below. Task results will be presented at the MediaEval 2017 Workshop, 13-15 September in Dublin, Ireland, co-located with CLEF 2017.
Deadline for regular registration: 1 May 2017
Late registration is possible after 1 May, but may narrow the window that you have to work with the data. See data release deadlines for the individual tasks at http://www.multimediaeval.org/mediaeval2017
MediaEval 2017 Tasks:
*Retrieving Diverse Social Images Task* Create an image retrieval system that is capable of returning diversified lists of results in response to complex queries.
*Emotional Impact of Movies Task* Design/implement a (multimodal) classifier to predict the emotion induced by a movie clip. This year we expand to detecting content that causes fear in kids.
*Predicting Media Interestingness Task* Design/implement an algorithm that predicts interesting frames or clips of a movie/video that help viewers decide which content to watch.
*Multimedia Satellite Task* Social images can be used to enrich events that are detected by remote sensing. Create an image retrieval system that retrieves images that depict flooding on the ground, and align them with evidence of flooding in satellite images.
*Medico: Medical Multimedia Task* Design/implement a classifier that is able to detect diseases in videos taken inside the intestines. Automatically generate appropriate captions.
*AcousticBrainz Genre Task: Content-based music genre recognition from multiple sources* Create a genre prediction system for music capable of leveraging labels from multiple information sources.
*C at MERATA: Querying Musical Scores with English Noun Phrases Task* Create a system which accepts as input a natural language phrase referring to a musical feature (e.g., ‘consecutive fifths’) and outputs a list of passages in a music score which contain that feature.
For more information see http://www.multimediaeval.org/mediaeval2017 and/or check out the report on the 2016 workshop:
M. Larson, M. Soleymani, G. Gravier, B. Ionescu, G.J.F. Jones. 2017. The Benchmarking Initiative for Multimedia Evaluation: MediaEval 2016. IEEE Multimedia, Vol. 24, No. 1, 93-96 http://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=7849098
If you have further questions, please contact: Martha Larson m.a.larson at tudelft.nl