{"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/domain-specific-face-synthesis-for-video-face","title":"Domain-Specific Face Synthesis for Video Face Recognition from a Single Sample Per Person","arxiv_id":"1801.01974","date":"2018-01-06","proceeding":null,"authors":["Fania Mokhayeri","Eric Granger","Guillaume-Alexandre Bilodeau"],"abstract":"The performance of still-to-video FR systems can decline significantly\nbecause faces captured in unconstrained operational domain (OD) over multiple\nvideo cameras have a different underlying data distribution compared to faces\ncaptured under controlled conditions in the enrollment domain (ED) with a still\ncamera. This is particularly true when individuals are enrolled to the system\nusing a single reference still. To improve the robustness of these systems, it\nis possible to augment the reference set by generating synthetic faces based on\nthe original still. However, without knowledge of the OD, many synthetic images\nmust be generated to account for all possible capture conditions. FR systems\nmay, therefore, require complex implementations and yield lower accuracy when\ntraining on many less relevant images. This paper introduces an algorithm for\ndomain-specific face synthesis (DSFS) that exploits the representative\nintra-class variation information available from the OD. Prior to operation, a\ncompact set of faces from unknown persons appearing in the OD is selected\nthrough clustering in the captured condition space. The domain-specific\nvariations of these face images are projected onto the reference stills by\nintegrating an image-based face relighting technique inside the 3D\nreconstruction framework. A compact set of synthetic faces is generated that\nresemble individuals of interest under the capture conditions relevant to the\nOD. In a particular implementation based on sparse representation\nclassification, the synthetic faces generated with the DSFS are employed to\nform a cross-domain dictionary that account for structured sparsity.\nExperimental results reveal that augmenting the reference gallery set of FR\nsystems using the proposed DSFS approach can provide a higher level of accuracy\ncompared to state-of-the-art approaches, with only a moderate increase in its\ncomputational complexity.","url_abs":"http://arxiv.org/abs/1801.01974v2","url_pdf":"http://arxiv.org/pdf/1801.01974v2.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":"domain-specific-face-synthesis-for-video-face","repo_url":"https://github.com/faniamokhayeri/DSFS","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"none","reach":null}],"tasks":[{"task_slug":"3d-reconstruction","task_name":"3D Reconstruction"},{"task_slug":"clustering","task_name":"Clustering"},{"task_slug":"face-generation","task_name":"Face Generation"},{"task_slug":"face-recognition","task_name":"Face Recognition"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":null,"mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}