{"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/travelgan-image-to-image-translation-by","title":"TraVeLGAN: Image-to-image Translation by Transformation Vector Learning","arxiv_id":"1902.09631","date":"2019-02-25","proceeding":"CVPR 2019 6","authors":["Matthew Amodio","Smita Krishnaswamy"],"abstract":"Interest in image-to-image translation has grown substantially in recent\nyears with the success of unsupervised models based on the cycle-consistency\nassumption. The achievements of these models have been limited to a particular\nsubset of domains where this assumption yields good results, namely homogeneous\ndomains that are characterized by style or texture differences. We tackle the\nchallenging problem of image-to-image translation where the domains are defined\nby high-level shapes and contexts, as well as including significant clutter and\nheterogeneity. For this purpose, we introduce a novel GAN based on preserving\nintra-domain vector transformations in a latent space learned by a siamese\nnetwork. The traditional GAN system introduced a discriminator network to guide\nthe generator into generating images in the target domain. To this two-network\nsystem we add a third: a siamese network that guides the generator so that each\noriginal image shares semantics with its generated version. With this new\nthree-network system, we no longer need to constrain the generators with the\nubiquitous cycle-consistency restraint. As a result, the generators can learn\nmappings between more complex domains that differ from each other by large\ndifferences - not just style or texture.","url_abs":"http://arxiv.org/abs/1902.09631v1","url_pdf":"http://arxiv.org/pdf/1902.09631v1.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":"travelgan-image-to-image-translation-by","repo_url":"https://github.com/KrishnaswamyLab/travelgan","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"tf","reach":null},{"paper_slug":"travelgan-image-to-image-translation-by","repo_url":"https://github.com/Medabid1/TravelGAN","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":null}],"tasks":[{"task_slug":"image-to-image-translation","task_name":"Image-to-Image Translation"},{"task_slug":"translation","task_name":"Translation"}],"methods":[{"method_slug":null,"method_name":null},{"method_slug":null,"method_name":null},{"method_slug":"convolution","method_name":"Convolution"},{"method_slug":null,"method_name":null},{"method_slug":"siamese-network","method_name":"Siamese Network"},{"method_slug":null,"method_name":"jackhenry"}],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"syntology_url":null,"atlas_url":"https://app.syntology.ai/?focus=1902.09631","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}