Data sets

Challenge protocol

The aim of this challenge is to develop relevant features and classification methods for recognizing social hand gestures. Participants have to classify the gestures independent of the variation/ condition (e.g. `gentle stroke' and `normal stroke' are both part of the same class). Participants can choose to work on one of the data sets or on both. The challenge procedure will contain standardized train-and-test protocols and error measures to allow for direct comparison. Each dataset will be divided into train and test sets. Participants will get a single opportunity to send in their labels for the test set and will receive their classification results and at the same time the test set labels will be released.

In order to get access to the data sets please register for the touch challenge by sending an email with you name(s) and affiliation(s) to touch.challenge@gmail.com.