The challenge data contains: audio, video and meta-data. The meta-data is composed of actor identity, age and gender. The participants are welcome to use any combination of modalities. 

2. Can scene information other than face information be used?

     Context analysis in FER is a hot topic. Participants can use scene/background/body pose etc. information along with the face information.

3. Which face and fiducial points detector have you used?

    For face detection, we found Zhu and Ramanan's mixture-of-parts based detector useful in our experiments. The authors have made public ally available an implementation of their method on: LINK. The tracking was performed using the Intraface tracker [LINK].

4. Can I use both train and validate data for learning my model?

    For evaluating a method on the Test set, data from both, Train and Val can be used for learning the model. 

5. Is the use of commercial face detector such as the Google Picasa OK?

    Any face detector whether commercial or academic can be used to participate in the challenge. The paper accompanying the challenge result submission should contain clear details of the detectors/libraries used.

6. Can I learn my model on both labelled train and unlabelled test data?

    No, the datasets are subject independent, test data is supposed to be used for testing purposes only.

7. Can I use external data for training along with the one provided?

    The participants are free to use external data for trainig along with the AFEW Train and Val partitions. However, this should be clearly discussed in the accompanying paper.

8. Will the review process be anyonomous?

    The review process is double-blind.

9. Is it necessary to participate in both the challenges?

    Participants can contribute to both or either of the challenges.

10. Mismtach between number of samples in a set (Train/Val) and aligned faces/features?

    For some sample, the face detector failed in the Train and Val sets. For these there are no aligned faces and features provided..