Cross Euclidean-to-Riemannian Metric Learning with Application to Face Recognition from Video release_6j2x3shm2baopbcw2akrm6jk7m

by Zhiwu Huang, Ruiping Wang, Shiguang Shan, Luc Van Gool, Xilin Chen

Entity Metadata (schema)

abstracts[] {'sha1': '80c28cfe8d22e82c923a42f7cb3397f121814fc7', 'content': 'Riemannian manifolds have been widely employed for video representations in\nvisual classification tasks including video-based face recognition. The success\nmainly derives from learning a discriminant Riemannian metric which encodes the\nnon-linear geometry of the underlying Riemannian manifolds. In this paper, we\npropose a novel metric learning framework to learn a distance metric across a\nEuclidean space and a Riemannian manifold to fuse the average appearance and\npattern variation of faces within one video. The proposed metric learning\nframework can handle three typical tasks of video-based face recognition:\nVideo-to-Still, Still-to-Video and Video-to-Video settings. To accomplish this\nnew framework, by exploiting typical Riemannian geometries for kernel\nembedding, we map the source Euclidean space and Riemannian manifold into a\ncommon Euclidean subspace, each through a corresponding high-dimensional\nReproducing Kernel Hilbert Space (RKHS). With this mapping, the problem of\nlearning a cross-view metric between the two source heterogeneous spaces can be\nexpressed as learning a single-view Euclidean distance metric in the target\ncommon Euclidean space. By learning information on heterogeneous data with the\nshared label, the discriminant metric in the common space improves face\nrecognition from videos. Extensive experiments on four challenging video face\ndatabases demonstrate that the proposed framework has a clear advantage over\nthe state-of-the-art methods in the three classical video-based face\nrecognition tasks.', 'mimetype': 'text/plain', 'lang': 'en'}
contribs[] {'index': 0, 'creator_id': None, 'creator': None, 'raw_name': 'Zhiwu Huang', 'given_name': None, 'surname': None, 'role': 'author', 'raw_affiliation': None, 'extra': None}
{'index': 1, 'creator_id': None, 'creator': None, 'raw_name': 'Ruiping Wang', 'given_name': None, 'surname': None, 'role': 'author', 'raw_affiliation': None, 'extra': None}
{'index': 2, 'creator_id': None, 'creator': None, 'raw_name': 'Shiguang Shan', 'given_name': None, 'surname': None, 'role': 'author', 'raw_affiliation': None, 'extra': None}
{'index': 3, 'creator_id': None, 'creator': None, 'raw_name': 'Luc Van Gool', 'given_name': None, 'surname': None, 'role': 'author', 'raw_affiliation': None, 'extra': None}
{'index': 4, 'creator_id': None, 'creator': None, 'raw_name': 'Xilin Chen', 'given_name': None, 'surname': None, 'role': 'author', 'raw_affiliation': None, 'extra': None}
ext_ids {'doi': None, 'wikidata_qid': None, 'isbn13': None, 'pmid': None, 'pmcid': None, 'core': None, 'arxiv': '1608.04200v2', 'jstor': None, 'ark': None, 'mag': None, 'doaj': None, 'dblp': None, 'oai': None, 'hdl': None}
files[] {'state': 'active', 'ident': 'wf77phbnqjfwndu4uf5fvn2iuu', 'revision': '458d0f79-39a3-4654-b20d-fdf64e7aca2a', 'redirect': None, 'extra': None, 'edit_extra': None, 'size': 928386, 'md5': 'dd2c09f004e115d4305983a44b07691b', 'sha1': '10e7dd3bbbfbc25661213155e0de1a9f043461a2', 'sha256': '369664dde36339b8f593cb47a5f1dcad990264017202dcc098a0c15a576593cf', 'urls': [{'url': '', 'rel': 'repository'}, {'url': '', 'rel': 'webarchive'}], 'mimetype': 'application/pdf', 'content_scope': None, 'release_ids': ['6j2x3shm2baopbcw2akrm6jk7m'], 'releases': None}
filesets []
language en
license_slug ARXIV-1.0
refs []
release_date 2017-01-06
release_stage submitted
release_type article
release_year 2017
title Cross Euclidean-to-Riemannian Metric Learning with Application to Face Recognition from Video
version v2
webcaptures []
work_id wgaobkaaknd2zpcsnp3tqovo3y

Extra Metadata (raw JSON)

arxiv.base_id 1608.04200
arxiv.categories ['cs.CV']