{"url":"/method/conditional-instance-normalization","slug":"conditional-instance-normalization","name":"Conditional Instance Normalization","full_name":"Conditional Instance Normalization","full_name_withheld":false,"description_markdown":"**Conditional Instance Normalization** is a normalization technique where all convolutional weights of a style transfer network are shared across many styles.  The goal of the procedure is transform\r\na layer’s activations $x$ into a normalized activation $z$ specific to painting style $s$. Building off\r\n[instance normalization](https://paperswithcode.com/method/instance-normalization), we augment the $\\gamma$ and $\\beta$ parameters so that they’re $N \\times C$ matrices, where $N$ is the number of styles being modeled and $C$ is the number of output feature maps. Conditioning on a style is achieved as follows:\r\n\r\n$$ z = \\gamma\\_{s}\\left(\\frac{x - \\mu}{\\sigma}\\right) + \\beta\\_{s}$$\r\n\r\nwhere $\\mu$ and $\\sigma$ are $x$’s mean and standard deviation taken across spatial axes and $\\gamma\\_{s}$ and $\\beta\\_{s}$ are obtained by selecting the row corresponding to $s$ in the $\\gamma$ and $\\beta$ matrices. One added benefit of this approach is that one can stylize a single image into $N$ painting styles with a single feed forward pass of the network with a batch size of $N$.","description_state":"present","introduced_year":null,"introduced_by":{"title":"A Learned Representation For Artistic Style","paper":"/paper/a-learned-representation-for-artistic-style","first_author":"Vincent Dumoulin","n_authors":3,"url_abs":null,"archive_paper_url":"https://paperswithcode.com/paper/a-learned-representation-for-artistic-style"},"source":{"url":"http://arxiv.org/abs/1610.07629v5","title":"A Learned Representation For Artistic Style","url_on_a_paper_host":true},"code_snippet_url":"https://github.com/kewellcjj/pytorch-multiple-style-transfer/blob/8a78000360cc36d3bbabdf838b91b6d12f88ae82/transformer_net.py#L46","code_snippet_url_on_a_code_host":true,"categories":[{"area":"General","area_id":"general","collection":"Normalization","url":"/methods/category/normalization","pwc_aliases":[]}],"n_papers_tagged":3,"archive_num_papers":3,"papers_newest_first":[{"paper":null,"title":"Multi-defect microscopy image restoration under limited data conditions","date":"2019-10-31","arxiv_id":"1910.14207","n_code_links":0,"syntology":null},{"paper":"/paper/exploring-the-structure-of-a-real-time","title":"Exploring the structure of a real-time, arbitrary neural artistic stylization network","date":"2017-05-18","arxiv_id":"1705.06830","n_code_links":20,"syntology":null},{"paper":"/paper/a-learned-representation-for-artistic-style","title":"A Learned Representation For Artistic Style","date":"2016-10-24","arxiv_id":"1610.07629","n_code_links":11,"syntology":null}],"papers_shown":3,"tasks":[{"task":"/task/data-augmentation","name":"Data Augmentation","papers":1},{"task":"/task/diversity","name":"Diversity","papers":1},{"task":null,"name":"Generative Adversarial Network","papers":1},{"task":"/task/image-restoration","name":"Image Restoration","papers":1},{"task":"/task/style-transfer","name":"Style Transfer","papers":1}],"tasks_shown":5,"n_tasks":5,"usage_by_year":[{"year":"2016","papers":1},{"year":"2017","papers":1},{"year":"2019","papers":1}],"row_source":"methods_table","archive":{"source":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","archive_url":"https://paperswithcode.com/method/conditional-instance-normalization"},"syntology_read_at":"2026-09-24T18:15:14+00:00"}