VisTa: Visual-contextual and Text-augmented Zero-shot Object-level OOD Detection | IEEE Conference Publication | IEEE Xplore