{"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/the-megaface-benchmark-1-million-faces-for","title":"The MegaFace Benchmark: 1 Million Faces for Recognition at Scale","arxiv_id":"1512.00596","date":"2015-12-02","proceeding":"CVPR 2016 6","authors":["Ira Kemelmacher-Shlizerman","Steve Seitz","Daniel Miller","Evan Brossard"],"abstract":"Recent face recognition experiments on a major benchmark LFW show stunning\nperformance--a number of algorithms achieve near to perfect score, surpassing\nhuman recognition rates. In this paper, we advocate evaluations at the million\nscale (LFW includes only 13K photos of 5K people). To this end, we have\nassembled the MegaFace dataset and created the first MegaFace challenge. Our\ndataset includes One Million photos that capture more than 690K different\nindividuals. The challenge evaluates performance of algorithms with increasing\nnumbers of distractors (going from 10 to 1M) in the gallery set. We present\nboth identification and verification performance, evaluate performance with\nrespect to pose and a person's age, and compare as a function of training data\nsize (number of photos and people). We report results of state of the art and\nbaseline algorithms. Our key observations are that testing at the million scale\nreveals big performance differences (of algorithms that perform similarly well\non smaller scale) and that age invariant recognition as well as pose are still\nchallenging for most. The MegaFace dataset, baseline code, and evaluation\nscripts, are all publicly released for further experimentations at:\nmegaface.cs.washington.edu.","url_abs":"http://arxiv.org/abs/1512.00596v1","url_pdf":"http://arxiv.org/pdf/1512.00596v1.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":[],"tasks":[{"task_slug":"face-recognition","task_name":"Face Recognition"}],"methods":[],"datasets_introduced":[{"slug":"megaface","name":"MegaFace","full_name":""}],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1512.00596","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}