Learning by Aligning: Visible-Infrared Person Re-identification using Cross-Modal Correspondences | IEEE Conference Publication | IEEE Xplore