2ND Workshop on learning with few or without annotated face, body and gesture data