Open AccessPosted Content
Prototype Memory for Large-scale Face Representation Learning.
TLDR
Prototype Memory as discussed by the authors uses a limited-size memory module for storing recent class prototypes and employs a set of algorithms to update it in appropriate way, which can be used with various loss functions, hard example mining algorithms and encoder architectures.Abstract:
Face representation learning using datasets with massive number of identities requires appropriate training methods. Softmax-based approach, currently the state-of-the-art in face recognition, in its usual "full softmax" form is not suitable for datasets with millions of persons. Several methods, based on the "sampled softmax" approach, were proposed to remove this limitation. These methods, however, have a set of disadvantages. One of them is a problem of "prototype obsolescence": classifier weights (prototypes) of the rarely sampled classes, receive too scarce gradients and become outdated and detached from the current encoder state, resulting in an incorrect training signals. This problem is especially serious in ultra-large-scale datasets. In this paper, we propose a novel face representation learning model called Prototype Memory, which alleviates this problem and allows training on a dataset of any size. Prototype Memory consists of the limited-size memory module for storing recent class prototypes and employs a set of algorithms to update it in appropriate way. New class prototypes are generated on the fly using exemplar embeddings in the current mini-batch. These prototypes are enqueued to the memory and used in a role of classifier weights for usual softmax classification-based training. To prevent obsolescence and keep the memory in close connection with encoder, prototypes are regularly refreshed, and oldest ones are dequeued and disposed. Prototype Memory is computationally efficient and independent of dataset size. It can be used with various loss functions, hard example mining algorithms and encoder architectures. We prove the effectiveness of the proposed model by extensive experiments on popular face recognition benchmarks.read more
Citations
More filters
Posted Content
STC speaker recognition systems for the NIST SRE 2021.
Anastasia Avdeeva,Aleksei Gusev,Igor Korsunov,Alexander Kozlov,Galina Lavrentyeva,Sergey Novoselov,Timur Pekhovsky,Andrey Shulipa,Alisa Vinogradova,Vladimir Volokhov,Evgeny Smirnov,Vasiliy Galyuk +11 more
TL;DR: In this paper, a number of diverse subsystems based on using deep neural networks as feature extractors were used for speaker verification filed in the NIST 2021 Speaker Recognition Evaluation for both fixed and open training conditions.
Book ChapterDOI
FaceMix: Transferring Local Regions for Data Augmentation in Face Recognition
TL;DR: FaceMix as mentioned in this paper is a flexible face-specific data augmentation technique that transfers a local area of an image to another image, and it can generate new images for a class, using face data from other classes, and these two modes also could be combined.
References
More filters
Proceedings ArticleDOI
RegularFace: Deep Face Recognition via Exclusive Regularization
TL;DR: The proposed method, named RegularFace, explicitly distances identities by penalizing the angle between an identity and its nearest neighbor, resulting in discriminative face representations, which is easy to implement and requires only a few lines of python code on modern deep learning frameworks.
Proceedings ArticleDOI
Co-Mining: Deep Face Recognition With Noisy Labels
TL;DR: A novel co-mining strategy that simultaneously uses the loss values as the cue to detect noisy labels, exchange the high-confidence clean faces to alleviate the errors accumulated issue caused by the sample-selection bias, and re-weight the predictedClean faces to make them dominate the discriminative model training in a mini-batch fashion.
Journal ArticleDOI
Mis-Classified Vector Guided Softmax Loss for Face Recognition
TL;DR: This paper develops a novel loss function, which adaptively emphasizes the mis-classified feature vectors to guide the discriminative feature learning and is the first attempt to inherit the advantages of feature margin and feature mining into a unified loss function.
Proceedings ArticleDOI
UniformFace: Learning Deep Equidistributed Representation for Face Recognition
Yueqi Duan,Jiwen Lu,Jie Zhou +2 more
TL;DR: A new supervision objective named uniform loss to learn deep equidistributed representations for face recognition is proposed, considering the class centers as like charges on the surface of hypersphere with inter-class repulsion, and minimize the total electric potential energy as the uniform loss.
Posted Content
Rethinking Feature Discrimination and Polymerization for Large-scale Recognition.
Yu Liu,Hongyang Li,Xiaogang Wang +2 more
TL;DR: The congenerous cosine (COCO) algorithm is proposed to simultaneously optimize the cosine similarity among data and inherits the softmax property to make inter-class features discriminative as well as shares the idea of class centroid in metric learning.