MegaFace: A Million Faces for Recognition at Scale

By Daniel Miller, Ira Kemelmacher-Shlizerman, S. M. Seitz. 2015

56 citations recorded by the source corpus.

Recent face recognition experiments on the LFW benchmark show that face recognition is performing stunningly well, surpassing human recognition rates. In this paper, we study face recognition at scale. Specifically, we have collected from Flickr a \textbf{Million} faces and evaluated state of the art face recognition algorithms on this dataset. We found that the performance of algorithms varies--while all perform great on LFW, once evaluated at scale recognition rates drop drastically for most algorithms. Interestingly, deep learning based approach by \cite{schroff2015facenet} performs much better, but still gets less robust at scale. We consider both verification and identification problems, and evaluate how pose affects recognition at scale. Moreover, we ran an extensive human study on Mechanical Turk to evaluate human recognition at scale, and report results. All the photos are creative commons photos and is released at \small{\url{this http URL}} for research and further experiments.

Subfields: Computer Vision and Pattern Recognition

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computer Vision and Pattern Recognition

Citations in the corpus

SphereFace: Deep Hypersphere Embedding for Face Recognition (2017, background): “MegaFace dataset [18] is a recently released testing benchmark with very challenging task to evaluate the performance of face recognition methods at the million scale of distractors.”
AdvHat: Real-World Adversarial Attack on ArcFace Face ID System (2019, methodology): “For example, the ArcFace (Insightface) solution is comparable with Microsoft and Google models in such challenges as Megaface [11, 12], NIST FRVT [13], and Trillion Pairs [14].”
UMDFaces: An annotated face dataset for training deep networks (2016, background): “The recently released Megaface challenge [19] might be the most difficult recognition (identification) benchmark currently.”
SphereFace Revived: Unifying Hyperspherical Face Recognition (2021, unknown): “MegaFace (distractor) [86] 690K 1M test”
Labeled Faces in the Wild: A Survey (2016, methodology): “Another new database designed to study large scale face recognition is MegaFace [67], a database of one million face images derived from the Yahoo 100 Million Flickr creative commons data set [98].”
A Comprehensive Study on Center Loss for Deep Face Recognition (2019, methodology): “Evaluation dataset and protocols MegaFace Challenge 1 (Miller et al. 2015) is released as a testing benchmark.”
A Discriminative Feature Learning Approach for Deep Face Recognition (2016, methodology): “4, extensive experiments are conducted on several public domain face datasets (LFW [15], YTF [38] and MegaFace Challenge [23]) to verify the effectiveness of the proposed approach.”
AgeDB: The First Manually Collected, In-the-Wild Age Database (2017, methodology): “In 2015 and 2016, the MegaFace database [18, 31] was introduced and extended, respectively.”
Frankenstein: Learning Deep Face Representations Using Small Data (2016, unknown): “6 million labeled face images [5], [31], [34], [52], [63].”