{"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/super-fan-integrated-facial-landmark","title":"Super-FAN: Integrated facial landmark localization and super-resolution of real-world low resolution faces in arbitrary poses with GANs","arxiv_id":"1712.02765","date":"2017-12-07","proceeding":"CVPR 2018 6","authors":["Adrian Bulat","Georgios Tzimiropoulos"],"abstract":"This paper addresses 2 challenging tasks: improving the quality of low\nresolution facial images and accurately locating the facial landmarks on such\npoor resolution images. To this end, we make the following 5 contributions: (a)\nwe propose Super-FAN: the very first end-to-end system that addresses both\ntasks simultaneously, i.e. both improves face resolution and detects the facial\nlandmarks. The novelty or Super-FAN lies in incorporating structural\ninformation in a GAN-based super-resolution algorithm via integrating a\nsub-network for face alignment through heatmap regression and optimizing a\nnovel heatmap loss. (b) We illustrate the benefit of training the two networks\njointly by reporting good results not only on frontal images (as in prior work)\nbut on the whole spectrum of facial poses, and not only on synthetic low\nresolution images (as in prior work) but also on real-world images. (c) We\nimprove upon the state-of-the-art in face super-resolution by proposing a new\nresidual-based architecture. (d) Quantitatively, we show large improvement over\nthe state-of-the-art for both face super-resolution and alignment. (e)\nQualitatively, we show for the first time good results on real-world low\nresolution images.","url_abs":"http://arxiv.org/abs/1712.02765v2","url_pdf":"http://arxiv.org/pdf/1712.02765v2.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":[],"tasks":[{"task_slug":"face-alignment","task_name":"Face Alignment"},{"task_slug":"face-hallucination","task_name":"Face Hallucination"},{"task_slug":"image-super-resolution","task_name":"Image Super-Resolution"},{"task_slug":"super-resolution","task_name":"Super-Resolution"}],"methods":[{"method_slug":"heatmap","method_name":"Heatmap"}],"datasets_introduced":[],"methods_introduced":[],"results":[{"leaderboard":"/sota/face-hallucination-on-ffhq-512-x-512-16x","task":"Face Hallucination","dataset":"FFHQ 512 x 512 - 16x upscaling","model":"Super-FAN","rank_in_archive_order":4,"of":4,"metrics":{"FID":"63.693","LPIPS":"0.4411","NIQE":"7.444"},"uses_additional_data":false},{"leaderboard":"/sota/image-super-resolution-on-ffhq-512-x-512-4x","task":"Image Super-Resolution","dataset":"FFHQ 512 x 512 - 4x upscaling","model":"Super-FAN","rank_in_archive_order":8,"of":8,"metrics":{"FED":"0.1416","FID":"14.811","LLE":"2.333","LPIPS":"0.2357","MS-SSIM":"0.913","NIQE":"8.719","PSNR":"25.463","SSIM":"0.729"},"uses_additional_data":false}],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1712.02765","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}