Unsupervised Object-Level Image-to-Image Translation Using Positional Attention Bi-Flow Generative Network | IEEE Journals & Magazine | IEEE Xplore