{"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/learning-from-millions-of-3d-scans-for-large","title":"Learning from Millions of 3D Scans for Large-scale 3D Face Recognition","arxiv_id":"1711.05942","date":"2017-11-16","proceeding":"CVPR 2018 6","authors":["Syed Zulqarnain Gilani","Ajmal Mian"],"abstract":"Deep networks trained on millions of facial images are believed to be closely\napproaching human-level performance in face recognition. However, open world\nface recognition still remains a challenge. Although, 3D face recognition has\nan inherent edge over its 2D counterpart, it has not benefited from the recent\ndevelopments in deep learning due to the unavailability of large training as\nwell as large test datasets. Recognition accuracies have already saturated on\nexisting 3D face datasets due to their small gallery sizes. Unlike 2D\nphotographs, 3D facial scans cannot be sourced from the web causing a\nbottleneck in the development of deep 3D face recognition networks and\ndatasets. In this backdrop, we propose a method for generating a large corpus\nof labeled 3D face identities and their multiple instances for training and a\nprotocol for merging the most challenging existing 3D datasets for testing. We\nalso propose the first deep CNN model designed specifically for 3D face\nrecognition and trained on 3.1 Million 3D facial scans of 100K identities. Our\ntest dataset comprises 1,853 identities with a single 3D scan in the gallery\nand another 31K scans as probes, which is several orders of magnitude larger\nthan existing ones. Without fine tuning on this dataset, our network already\noutperforms state of the art face recognition by over 10%. We fine tune our\nnetwork on the gallery set to perform end-to-end large scale 3D face\nrecognition which further improves accuracy. Finally, we show the efficacy of\nour method for the open world face recognition problem.","url_abs":"http://arxiv.org/abs/1711.05942v3","url_pdf":"http://arxiv.org/pdf/1711.05942v3.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":"learning-from-millions-of-3d-scans-for-large","repo_url":"https://github.com/huyhieupham/3D-Face-Recognition","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"none","reach":null}],"tasks":[{"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}