FACSCaps: Pose-Independent Facial Action Coding with Capsules | IEEE Conference Publication | IEEE Xplore