{"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/perceptual-adversarial-networks-for-image-to","title":"Perceptual Adversarial Networks for Image-to-Image Transformation","arxiv_id":"1706.09138","date":"2017-06-28","proceeding":null,"authors":["Chaoyue Wang","Chang Xu","Chaohui Wang","DaCheng Tao"],"abstract":"In this paper, we propose a principled Perceptual Adversarial Networks (PAN)\nfor image-to-image transformation tasks. Unlike existing application-specific\nalgorithms, PAN provides a generic framework of learning mapping relationship\nbetween paired images (Fig. 1), such as mapping a rainy image to its de-rained\ncounterpart, object edges to its photo, semantic labels to a scenes image, etc.\nThe proposed PAN consists of two feed-forward convolutional neural networks\n(CNNs), the image transformation network T and the discriminative network D.\nThrough combining the generative adversarial loss and the proposed perceptual\nadversarial loss, these two networks can be trained alternately to solve\nimage-to-image transformation tasks. Among them, the hidden layers and output\nof the discriminative network D are upgraded to continually and automatically\ndiscover the discrepancy between the transformed image and the corresponding\nground-truth. Simultaneously, the image transformation network T is trained to\nminimize the discrepancy explored by the discriminative network D. Through the\nadversarial training process, the image transformation network T will\ncontinually narrow the gap between transformed images and ground-truth images.\nExperiments evaluated on several image-to-image transformation tasks (e.g.,\nimage de-raining, image inpainting, etc.) show that the proposed PAN\noutperforms many related state-of-the-art methods.","url_abs":"http://arxiv.org/abs/1706.09138v2","url_pdf":"http://arxiv.org/pdf/1706.09138v2.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":"perceptual-adversarial-networks-for-image-to","repo_url":"https://github.com/DLHacks/pix2pix_PAN","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"unanswered"}},{"paper_slug":"perceptual-adversarial-networks-for-image-to","repo_url":"https://github.com/nekitmm/starnet","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"tf","reach":{"status":"unanswered"}}],"tasks":[{"task_slug":"image-inpainting","task_name":"Image Inpainting"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1706.09138","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}