SphereFace: Deep Hypersphere Embedding for Face Recognition

CVPR 2017 Weiyang LiuYandong WenZhiding YuMing LiBhiksha RajLe Song

This paper addresses deep face recognition (FR) problem under open-set protocol, where ideal face features are expected to have smaller maximal intra-class distance than minimal inter-class distance under a suitably chosen metric space. However, few existing algorithms can effectively achieve this criterion... (read more)

PDF Abstract

Results from the Paper


TASK DATASET MODEL METRIC NAME METRIC VALUE GLOBAL RANK RESULT BENCHMARK
Face Verification CK+ SphereFace Accuracy 93.80 # 4
Face Verification Labeled Faces in the Wild SphereFace Accuracy 99.42% # 12
Face Identification MegaFace SphereFace (3-patch ensemble) Accuracy 75.766% # 4
Face Identification MegaFace SphereFace (single model) Accuracy 72.729% # 6
Face Verification MegaFace SphereFace (single model) Accuracy 85.561% # 8
Face Verification MegaFace SphereFace (3-patch ensemble) Accuracy 89.142% # 6
Face Verification YouTube Faces DB SphereFace Accuracy 95.0% # 9

Results from Other Papers


TASK DATASET MODEL METRIC NAME METRIC VALUE GLOBAL RANK SOURCE PAPER COMPARE
Face Verification Oulu-CASIA SphereFace Accuracy 95.50 # 4
Face Identification Trillion Pairs Dataset A-Softmax Accuracy 43.89 # 4
Face Verification Trillion Pairs Dataset A-Softmax Accuracy 43.76 # 4

Methods used in the Paper