{"about":{"site":"https://codewithpapers.app","non_affiliation":"Code with Papers and Syntology are not affiliated with, endorsed by, or sponsored by Papers with Code, Meta, or the pwc-archive mirror.","licence":"CC BY-SA 4.0","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","attribution":"https://codewithpapers.app/attribution","modified":"archive material modified by Syntology; see the attribution page"},"url":"/paper/cr-gan-learning-complete-representations-for","title":"CR-GAN: Learning Complete Representations for Multi-view Generation","arxiv_id":"1806.11191","date":"2018-06-28","proceeding":null,"authors":["Yu Tian","Xi Peng","Long Zhao","Shaoting Zhang","Dimitris N. Metaxas"],"abstract":"Generating multi-view images from a single-view input is an essential yet\nchallenging problem. It has broad applications in vision, graphics, and\nrobotics. Our study indicates that the widely-used generative adversarial\nnetwork (GAN) may learn \"incomplete\" representations due to the single-pathway\nframework: an encoder-decoder network followed by a discriminator network. We\npropose CR-GAN to address this problem. In addition to the single\nreconstruction path, we introduce a generation sideway to maintain the\ncompleteness of the learned embedding space. The two learning pathways\ncollaborate and compete in a parameter-sharing manner, yielding considerably\nimproved generalization ability to \"unseen\" dataset. More importantly, the\ntwo-pathway framework makes it possible to combine both labeled and unlabeled\ndata for self-supervised learning, which further enriches the embedding space\nfor realistic generations. The experimental results prove that CR-GAN\nsignificantly outperforms state-of-the-art methods, especially when generating\nfrom \"unseen\" inputs in wild conditions.","url_abs":"http://arxiv.org/abs/1806.11191v1","url_pdf":"http://arxiv.org/pdf/1806.11191v1.pdf","source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","row_kind":"abstracts"},"code_links":[{"paper_slug":"cr-gan-learning-complete-representations-for","repo_url":"https://github.com/bluer555/CR-GAN","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok"}}],"tasks":[{"task_slug":"decoder","task_name":"Decoder"},{"task_slug":null,"task_name":"Generative Adversarial Network"},{"task_slug":"self-supervised-learning","task_name":"Self-Supervised Learning"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1806.11191","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}