Robust face alignment with cascaded coarse-to-fine auto-encoder network | IEEE Conference Publication | IEEE Xplore