Vision-Language Pre-Training with Triple Contrastive Learning | IEEE Conference Publication | IEEE Xplore