Arcface: additive angular margin loss for deep face recognition

Abstract

One of the main challenges in feature learning usingDeep Convolutional Neural Networks (DCNNs) for large-scale face recognition is the design of appropriate loss func-tions that enhance discriminative power. Centre loss pe-nalises the distance between the deep features and their cor-responding class centres in the Euclidean space to achieveintra-class compactness. SphereFace assumes that the lin-ear transformation matrix in the last fully connected layercan be used as a representation of the class centres in anangular space and penalises the angles between the deepfeatures and their corresponding weights in a multiplicativeway. Recently, a popular line of research is to incorporatemargins in well-established loss functions in order to max-imise face class separability. In this paper, we propose anAdditive Angular Margin Loss (ArcFace) to obtain highlydiscriminative features for face recognition. The proposedArcFace has a clear geometric interpretation due to the ex-act correspondence to the geodesic distance on the hyper-sphere. We present arguably the most extensive experimen-tal evaluation of all the recent state-of-the-art face recog-nition methods on over 10 face recognition benchmarks in-cluding a new large-scale image database with trillion levelof pairs and a large-scale video dataset. We show that Ar-cFace consistently outperforms the state-of-the-art and canbe easily implemented with negligible computational over-head. We release all refined training data, training codes,pre-trained models and training logs1, which will help re-produce the results in this paper.