Invitation to the MediaEval 2019 multimodal speaker detection challenge.

We are running a MediaEval Challenge again this year for estimating speaking status.
Rather than using audio, the challenge involves using accelerometer data extracted from a
single body worn tri-axial accelerometer, and also video.

The idea is to leverage the body movements (e.g gestures) as a proxy for speaking status
rather than vocal chord vibration. Prior work shows that this is a viable option. We aim
to draw on the research community expertise to improve the performance further and
exploit interdisciplinary knowledge to understand the nature of the problem better.

We welcome you to apply your latest approaches on our challenge. Baseline code is also
available to get you started.

More details about the challenge can be found here:
http://multimediaeval.org/mediaeval2019/speakerturns/index.html

Task Schedule

Data release: 31 May
Run submission: 20 September
Results returned: 23 September

Working Notes paper deadline: 20 September

MediaEval 2019 Workshop (in France, near Nice): 27-29 October 2019