FacER: Contrastive Attention based Expression Recognition via Smartphone Earpiece Speaker | IEEE Conference Publication | IEEE Xplore