AVEC 2017 menu

Challenge data

For the challenge we will use a training and test set derived from the GEMEP database. We will refer to this dataset as the GEMEP-FERA dataset. The training partition is now available from an online-accessible web-database. The test partition will be made available for one working week only, which will be the week before the submission deadline.

For the AU recognition sub-challenge, participants are encouraged to enhance their training data with other freely available datasets, such as the MMI-Facial Expression Database (available online) and the Cohn-Kanade database (available through traditional mail).

For issues regarding the FACS coding of the data, please contact marc.mehu@unige.ch