ChairGest is an open challenge. You can submit a request to download the data and then start working on your algorithm. Once you are satisfied with your result, you can submit your algorithm for final evaluation by the organizers and, if you agree, your score will be included in the scoreboard page.


The following paper describes the corpus and and challenge in details:
Ruffieux, S. et al. 2013. ChAirGest: A Challenge for Multimodal Mid-Air Gesture Recognition for Close HCI. Proceedings of the 15th ACM on International conference on multimodal interaction - ICMI ’13(Sydney, Australia, 2013), 483–488. DOI=10.1145/2522848.2532590. [BibTex]


The proposed challenge targets two main goals: optimize the temporal segmentation and the recognition of one hand/arm gestures. The process of temporal gesture segmentation (also called gesture spotting) remains an important and challenging task in the domain; it consists in segmenting the start and end points of a gesture in a continuous sequence. The process of gesture recognition consists in correctly identifying a pre-segmented gesture within a known vocabulary. Therefore the task of the challenge can be resumed as follows:

Spot and recognize the gestures using the desired set of sensor(s) independently of subjects