Range Loss for Deep Face Recognition with Long-Tailed Training Data

By Xiao Zhang, Zhiyuan Fang, Yandong Wen, Zhifeng Li, Yu Qiao. 2016

472 citations recorded by the source corpus.

Deep convolutional neural networks have achieved significant improvements on face recognition task due to their ability to learn highly discriminative features from tremendous amounts of face images. Many large scale face datasets exhibit long-tail distribution where a small number of entities (persons) have large number of face images while a large number of persons only have very few face samples (long tail). Most of the existing works alleviate this problem by simply cutting the tailed data and only keep identities with enough number of examples. Unlike these work, this paper investigated how long-tailed data impact the training of face CNNs and develop a novel loss function, called range loss, to effectively utilize the tailed data in training process. More specifically, range loss is designed to reduce overall intrapersonal variations while enlarge interpersonal differences simultaneously. Extensive experiments on two face recognition benchmarks, Labeled Faces in the Wild (LFW) [11] and YouTube Faces (YTF) [33], demonstrate the effectiveness of the proposed range loss in overcoming the long tail effect, and show the good generalization ability of the proposed methods.

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

Long-Tailed Class Incremental Learning (2022, methodology): “Learning from long-tailed distributions has been approach by re-sampling [11,15] or re-weighting [21,36,48] head classes and tail classes to learn a balanced classifier.”
Large-Scale Long-Tailed Recognition in an Open World (2019, methodology): “Compared to the plain model [18] and range loss [59], the performance of our approach changes steadily as the open-set threshold rises.”
Decoupling Representation and Classifier for Long-Tailed Recognition (2019, methodology): “, 2017), Range Loss (Zhang et al., 2017), FSLwF (Gidaris & Komodakis, 2018) and OLTR (Liu et al.”
Class-Balanced Loss Based on Effective Number of Samples (2019, background): “A number of recent studies have aimed to alleviate the challenge of long-tailed training data [3, 32, 17, 42, 44, 12, 48, 45].”
CosFace: Large Margin Cosine Loss for Deep Face Recognition (2018, background): “Other loss functions [47, 6, 4, 3] based on contrastive loss or center loss also demonstrate the performance on enhancing discrimination.”
Meta-Weight-Net: Learning an Explicit Mapping For Sample Weighting (2019, methodology): “The metric learning based methods have also been developed to effectively exploit the tailed data to improve the generalization ability, e.g., triple-header loss [55] and range loss [56].”
AdaCos: Adaptively Scaling Cosine Logits for Effectively Learning Deep Face Representations (2019, background): “Taking advantages of these works, center loss [42] and range loss [44] were proposed to reduce intra-class variations via minimizing distances within each class [2].”
Rethinking the Value of Labels for Improving Class-Imbalanced Learning (2020, background): “Other learning paradigms, including transfer learning [33, 54], metric learning [55, 58], and meta-learning [1, 45], have also been explored.”
Low-Resolution Face Recognition in the Wild via Selective Knowledge Distillation (2018, background): “After that, various loss functions have been proposed for training face recognition CNNs, such as triplet loss [3], [4], center loss [15] and range loss [17].”
DeepSMOTE: Fusing Deep Learning and SMOTE for Imbalanced Data (2021, background): “Very interesting solutions to this problem are based on adaptation of the loss function in deep neural networks, such as equalization loss [78], hubless loss [79], and range loss [80].”
Solving Long-tailed Recognition with Deep Realistic Taxonomic Classifier (2020, unknown): “ 0.79 0.75 0.78 hier. acc. 0.92 0.91 0.89 0.90 leaf freq. 0.71 0.56 0.48 0.54 leaf acc. 0.76 0.67 0.60 0.64 Table 4: Results on ImageNet-LT. Method CPB FSLwF [26] 0.28 Focal Loss [44] 0.31 Range Loss [62] 0.31 Lifted Loss [54] 0.31 OLTR [46] 0.36 Softmax 0.35 NCM [38] 0.36 cRT [38] 0.42 ˝-norm [38] ”
Long-tail learning via logit adjustment (2020, background): “…decision threshold (Fawcett & Provost, 1996; Collell et al., 2016) or weights (Kim & Kim, 2019; Kang et al., 2020); or (iii) the training procedure of a model, for example by modifying the loss function (Zhang et al., 2017; Cui et al., 2019; Cao et al., 2019; Tan et al., 2020; Jamal et al., 2020).”
Deep Imbalanced Learning for Face Recognition and Attribute Prediction (2018, methodology): “The methods of minority class augmentation in image space [47] and feature space [48] obtain sub-optimal results, while the range loss [51] and center invariant loss [49] are not as discriminative as our CLMLE.”
Feature Transfer Learning for Face Recognition With Under-Represented Data (2019, background): “18 RangeLoss [50] 99 .”