{"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/deep-learning-face-representation-from-1","title":"Deep Learning Face Representation from Predicting 10,000 Classes","arxiv_id":null,"date":"2014-01-01","proceeding":null,"authors":["Yi Sun","Xiaogang Wang","Xiaoou Tang"],"abstract":"This paper proposes to learn a set of high-level feature\r\nrepresentations through deep learning, referred to as Deep\r\nhidden IDentity features (DeepID), for face verification.\r\nWe argue that DeepID can be effectively learned through\r\nchallenging multi-class face identification tasks, whilst they\r\ncan be generalized to other tasks (such as verification) and\r\nnew identities unseen in the training set. Moreover, the\r\ngeneralization capability of DeepID increases as more face\r\nclasses are to be predicted at training. DeepID features\r\nare taken from the last hidden layer neuron activations of\r\ndeep convolutional networks (ConvNets). When learned\r\nas classifiers to recognize about 10, 000 face identities in\r\nthe training set and configured to keep reducing the neuron\r\nnumbers along the feature extraction hierarchy, these deep\r\nConvNets gradually form compact identity-related features\r\nin the top layers with only a small number of hidden\r\nneurons. The proposed features are extracted from various\r\nface regions to form complementary and over-complete\r\nrepresentations. Any state-of-the-art classifiers can be\r\nlearned based on these high-level representations for face\r\nverification. 97.45% verification accuracy on LFW is\r\nachieved with only weakly aligned faces","url_abs":"http://mmlab.ie.cuhk.edu.hk/pdf/YiSun_CVPR14.pdf","url_pdf":"http://mmlab.ie.cuhk.edu.hk/pdf/YiSun_CVPR14.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":"deep-learning-face-representation-from-1","repo_url":"https://github.com/Recognito-Vision/Android-FaceRecognition-FaceLivenessDetection","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":0,"framework":"none","reach":{"status":"ok"}},{"paper_slug":"deep-learning-face-representation-from-1","repo_url":"https://github.com/Ruoyiran/DeepID","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":0,"framework":"tf","reach":{"status":"ok","spdx":"GPL-3.0"}},{"paper_slug":"deep-learning-face-representation-from-1","repo_url":"https://github.com/mindspore-ai/models/tree/master/research/cv/DeepID","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":0,"framework":"mindspore","reach":null},{"paper_slug":"deep-learning-face-representation-from-1","repo_url":"https://github.com/serengil/deepface","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":0,"framework":"tf","reach":{"status":"ok","spdx":"MIT"}}],"tasks":[{"task_slug":"deep-learning","task_name":"Deep Learning"},{"task_slug":"face-identification","task_name":"Face Identification"},{"task_slug":"face-verification","task_name":"Face Verification"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[{"leaderboard":"/sota/face-verification-on-labeled-faces-in-the","task":"Face Verification","dataset":"Labeled Faces in the Wild","model":"DeepID","rank_in_archive_order":6,"of":7,"metrics":{"Accuracy":"97.05%"},"uses_additional_data":false}],"syntology":{"atlas_url":null,"mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}