{"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/rankiqa-learning-from-rankings-for-no","title":"RankIQA: Learning from Rankings for No-reference Image Quality Assessment","arxiv_id":"1707.08347","date":"2017-07-26","proceeding":"ICCV 2017 10","authors":["Xialei Liu","Joost Van de Weijer","Andrew D. Bagdanov"],"abstract":"We propose a no-reference image quality assessment (NR-IQA) approach that\nlearns from rankings (RankIQA). To address the problem of limited IQA dataset\nsize, we train a Siamese Network to rank images in terms of image quality by\nusing synthetically generated distortions for which relative image quality is\nknown. These ranked image sets can be automatically generated without laborious\nhuman labeling. We then use fine-tuning to transfer the knowledge represented\nin the trained Siamese Network to a traditional CNN that estimates absolute\nimage quality from single images. We demonstrate how our approach can be made\nsignificantly more efficient than traditional Siamese Networks by forward\npropagating a batch of images through a single network and backpropagating\ngradients derived from all pairs of images in the batch. Experiments on the\nTID2013 benchmark show that we improve the state-of-the-art by over 5%.\nFurthermore, on the LIVE benchmark we show that our approach is superior to\nexisting NR-IQA techniques and that we even outperform the state-of-the-art in\nfull-reference IQA (FR-IQA) methods without having to resort to high-quality\nreference images to infer IQA.","url_abs":"http://arxiv.org/abs/1707.08347v1","url_pdf":"http://arxiv.org/pdf/1707.08347v1.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":"rankiqa-learning-from-rankings-for-no","repo_url":"https://github.com/xialeiliu/RankIQA","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"none","reach":{"status":"ok","spdx":"MIT"}},{"paper_slug":"rankiqa-learning-from-rankings-for-no","repo_url":"https://github.com/YunanZhu/Pytorch-TestRankIQA","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"unanswered"}}],"tasks":[{"task_slug":"image-quality-assessment","task_name":"Image Quality Assessment"},{"task_slug":"no-reference-image-quality-assessment","task_name":"No-Reference Image Quality Assessment"}],"methods":[{"method_slug":"siamese-network","method_name":"Siamese Network"}],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1707.08347","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}