{"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/loss-functions-for-neural-networks-for-image","title":"Loss Functions for Neural Networks for Image Processing","arxiv_id":"1511.08861","date":"2015-11-28","proceeding":null,"authors":["Hang Zhao","Orazio Gallo","Iuri Frosio","Jan Kautz"],"abstract":"Neural networks are becoming central in several areas of computer vision and\nimage processing and different architectures have been proposed to solve\nspecific problems. The impact of the loss layer of neural networks, however,\nhas not received much attention in the context of image processing: the default\nand virtually only choice is L2. In this paper, we bring attention to\nalternative choices for image restoration. In particular, we show the\nimportance of perceptually-motivated losses when the resulting image is to be\nevaluated by a human observer. We compare the performance of several losses,\nand propose a novel, differentiable error function. We show that the quality of\nthe results improves significantly with better loss functions, even when the\nnetwork architecture is left unchanged.","url_abs":"http://arxiv.org/abs/1511.08861v3","url_pdf":"http://arxiv.org/pdf/1511.08861v3.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":"loss-functions-for-neural-networks-for-image","repo_url":"https://github.com/NVlabs/PL4NN","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":0,"framework":"caffe2","reach":{"status":"ok"}},{"paper_slug":"loss-functions-for-neural-networks-for-image","repo_url":"https://github.com/kornia/kornia","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":0,"framework":"pytorch","reach":null}],"tasks":[{"task_slug":"image-restoration","task_name":"Image Restoration"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1511.08861","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}