ArcFace: Additive Angular Margin Loss for Deep Face Recognition
15 Jun 2019-pp 4690-4699
TL;DR: This paper presents arguably the most extensive experimental evaluation against all recent state-of-the-art face recognition methods on ten face recognition benchmarks, and shows that ArcFace consistently outperforms the state of the art and can be easily implemented with negligible computational overhead.
Abstract: One of the main challenges in feature learning using Deep Convolutional Neural Networks (DCNNs) for large-scale face recognition is the design of appropriate loss functions that can enhance the discriminative power. Centre loss penalises the distance between deep features and their corresponding class centres in the Euclidean space to achieve intra-class compactness. SphereFace assumes that the linear transformation matrix in the last fully connected layer can be used as a representation of the class centres in the angular space and therefore penalises the angles between deep features and their corresponding weights in a multiplicative way. Recently, a popular line of research is to incorporate margins in well-established loss functions in order to maximise face class separability. In this paper, we propose an Additive Angular Margin Loss (ArcFace) to obtain highly discriminative features for face recognition. The proposed ArcFace has a clear geometric interpretation due to its exact correspondence to geodesic distance on a hypersphere. We present arguably the most extensive experimental evaluation against all recent state-of-the-art face recognition methods on ten face recognition benchmarks which includes a new large-scale image database with trillions of pairs and a large-scale video dataset. We show that ArcFace consistently outperforms the state of the art and can be easily implemented with negligible computational overhead. To facilitate future research, the code has been made available.
Citations
More filters
TL;DR: CIDER jointly optimizes two losses to promote strong ID-OOD separability: a dispersion loss that promotes large angular distances among different class prototypes, and a compactness loss that encourages samples to be close to their class prototypes.
Abstract: Out-of-distribution (OOD) detection is a critical task for reliable machine learning. Recent advances in representation learning give rise to developments in distance-based OOD detection, where testing samples are detected as OOD if they are relatively far away from the centroids or prototypes of in-distribution (ID) classes. However, prior methods directly take off-the-shelf loss functions that suffice for classifying ID samples, but are not optimally designed for OOD detection. In this paper, we propose CIDER , a simple and effective representation learning framework by exploiting hyperspherical embeddings for OOD detection. CIDER jointly optimizes two losses to promote strong ID-OOD separability: (1) a dispersion loss that promotes large angular distances among different class prototypes, and (2) a compactness loss that encourages samples to be close to their class prototypes. We show that CIDER is effective under various settings and establishes state-of-the-art performance. On a hard OOD detection task CIFAR-100 vs. CIFAR-10, our method substantially improves the AUROC by 14 . 20% compared to the embeddings learned by the cross-entropy loss.
7 citations
01 Mar 2020
TL;DR: This work introduces a new dataset, named EDGE20, that can be used in addressing the problems of pedestrian detection, face detection, and face recognition in images captured using trail cameras under the VIS and NIR spectra.
Abstract: Surveillance-related datasets that have been released in recent years focus only on one specific problem at a time (e.g., pedestrian detection, face detection, or face recognition), while most of them were collected using visible spectrum (VIS) cameras. Even though some cross-spectral datasets were presented in the past, they were acquired in a constrained setup, which limited the performance of methods for the aforementioned problems under a cross-spectral setting. This work introduces a new dataset, named EDGE20, that can be used in addressing the problems of pedestrian detection, face detection, and face recognition in images captured using trail cameras under the VIS and NIR spectra. Data acquisition was performed in an outdoor environment, during both day and night, under unconstrained acquisition conditions. The collection of images is accompanied by a rich set of annotations, consisting of person and facial bounding boxes, unique subject identifiers, and labels that characterize facial images as frontal, profile, or back faces. Moreover, the performance of several state-of-the-art methods was evaluated for each of the scenarios covered by our dataset. The baseline results we obtained highlight the difficulty of current methods in the tasks of cross-spectral pedestrian detection, face detection, and face recognition due to unconstrained conditions, including low resolution, pose variation, illumination variation, occlusions, and motion blur.
7 citations
Cites background or methods from "ArcFace: Additive Angular Margin Lo..."
...Cosine similarity is used to measure the similarity of two feature vectors generated by ArcFace....
[...]
...To establish a baseline for face recognition, we employed ArcFace [6]....
[...]
...In Section 5, baseline experimental results from state-of-the-art methods for pedestrian detection [22], face detection [24], and face recognition [6] are reported....
[...]
...The evaluation results of ArcFace on EDGE20 are shown in Table....
[...]
...We trained ArcFace on the Deep Glint dataset [5]....
[...]
01 Jun 2021
TL;DR: In this paper, a small amount of unlabeled data with sufficient diversity can lead to an appreciable gain in recognition performance and outperform the supervised baseline when combined with less than half of the labeled data.
Abstract: In recent years, significant progress has been made in face recognition, which can be partially attributed to the availability of large-scale labeled face datasets. However, since the faces in these datasets usually contain limited degree and types of variation, the resulting trained models generalize poorly to more realistic unconstrained face datasets. While collecting labeled faces with larger variations could be helpful, it is practically infeasible due to privacy and labor cost. In comparison, it is easier to acquire a large number of unlabeled faces from different domains, which could be used to regularize the learning of face representations. We present an approach to use such unlabeled faces to learn generalizable face representations, where we assume neither the access to identity labels nor domain labels for unlabeled images. Experimental results on unconstrained datasets show that a small amount of unlabeled data with sufficient diversity can (i) lead to an appreciable gain in recognition performance and (ii) outperform the supervised baseline when combined with less than half of the labeled data. Compared with the state-of-the-art face recognition methods, our method further improves their performance on challenging benchmarks, such as IJBB, IJB-C and IJB-S.
7 citations
TL;DR: The implicit and explicit feature purification mechanism can produce facial-feature embeddings that preserve identity information as much as possible and are insensitive to age variations, and it tends to generate low-rank, yet high-dimensional, representations for age-invariant face recognition.
Abstract: This paper presents a new method, named implicit and explicit feature purification (IEFP), for age-invariant face recognition. Facial features extracted from a face image contain the information about the identity, age, and other attributes. For age-invariant face recognition, it is important to remove the irrelevant information, and retain the identity information only, in the facial features. Through the two proposed feature purification mechanisms, our framework can produce facial-feature embeddings that preserve identity information as much as possible and are insensitive to age variations. Specifically, on the one hand, a special network module is devised to implicitly purify the original facial features obtained from a face encoder. On the other hand, to obtain purer facial feature representations for age-invariant face recognition, irrelevant information within the implicitly purified features, such as the age, is further removed. This is realized by using a regularizer, based on information theory, to explicitly minimize the correlation between identity-related features and age-related features. Comprehensive ablation studies show that these two feature purification schemes can work independently, as well as collaboratively, to achieve better performance. Extensive evaluations on several benchmark data sets show that the IEFP method is on par with those competitors learned on far more favorable training samples, and it achieves the best performance in a fair comparison. Furthermore, we provide mathematical interpretation to explain the effectiveness of our approach, and find that it tends to generate low-rank, yet high-dimensional, representations for age-invariant face recognition.
7 citations
TL;DR: Zhang et al. as discussed by the authors proposed to generate realistic and identity-preserving face video from a single face image by reconstructing 3D face dynamics, which can serve as a strong prior knowledge for guiding highly realistic face video generation.
Abstract: We present a versatile model, FaceAnime, for various video generation tasks from still images. Video generation from a single face image is an interesting problem and usually tackled by utilizing Generative Adversarial Networks (GANs) to integrate information from the input face image and a sequence of sparse facial landmarks. However, the generated face images usually suffer from quality loss, image distortion, identity change, and expression mismatching due to the weak representation capacity of the facial landmarks. In this paper, we propose to “imagine” a face video from a single face image according to the reconstructed 3D face dynamics, aiming to generate a realistic and identity-preserving face video, with precisely predicted pose and facial expression. The 3D dynamics reveal changes of the facial expression and motion, and can serve as a strong prior knowledge for guiding highly realistic face video generation. In particular, we explore face video prediction and exploit a well-designed 3D dynamic prediction network to predict a 3D dynamic sequence for a single face image. The 3D dynamics are then further rendered by the sparse texture mapping algorithm to recover structural details and sparse textures for generating face frames. Our model is versatile for various AR/VR and entertainment applications, such as face video retargeting and face video prediction. Superior experimental results have well demonstrated its effectiveness in generating high-fidelity, identity-preserving, and visually pleasant face video clips from a single source face image.
7 citations
References
More filters
27 Jun 2016
TL;DR: In this article, the authors proposed a residual learning framework to ease the training of networks that are substantially deeper than those used previously, which won the 1st place on the ILSVRC 2015 classification task.
Abstract: Deeper neural networks are more difficult to train. We present a residual learning framework to ease the training of networks that are substantially deeper than those used previously. We explicitly reformulate the layers as learning residual functions with reference to the layer inputs, instead of learning unreferenced functions. We provide comprehensive empirical evidence showing that these residual networks are easier to optimize, and can gain accuracy from considerably increased depth. On the ImageNet dataset we evaluate residual nets with a depth of up to 152 layers—8× deeper than VGG nets [40] but still having lower complexity. An ensemble of these residual nets achieves 3.57% error on the ImageNet test set. This result won the 1st place on the ILSVRC 2015 classification task. We also present analysis on CIFAR-10 with 100 and 1000 layers. The depth of representations is of central importance for many visual recognition tasks. Solely due to our extremely deep representations, we obtain a 28% relative improvement on the COCO object detection dataset. Deep residual nets are foundations of our submissions to ILSVRC & COCO 2015 competitions1, where we also won the 1st places on the tasks of ImageNet detection, ImageNet localization, COCO detection, and COCO segmentation.
123,388 citations
Journal Article•
TL;DR: It is shown that dropout improves the performance of neural networks on supervised learning tasks in vision, speech recognition, document classification and computational biology, obtaining state-of-the-art results on many benchmark data sets.
Abstract: Deep neural nets with a large number of parameters are very powerful machine learning systems. However, overfitting is a serious problem in such networks. Large networks are also slow to use, making it difficult to deal with overfitting by combining the predictions of many different large neural nets at test time. Dropout is a technique for addressing this problem. The key idea is to randomly drop units (along with their connections) from the neural network during training. This prevents units from co-adapting too much. During training, dropout samples from an exponential number of different "thinned" networks. At test time, it is easy to approximate the effect of averaging the predictions of all these thinned networks by simply using a single unthinned network that has smaller weights. This significantly reduces overfitting and gives major improvements over other regularization methods. We show that dropout improves the performance of neural networks on supervised learning tasks in vision, speech recognition, document classification and computational biology, obtaining state-of-the-art results on many benchmark data sets.
33,597 citations
Proceedings Article•
06 Jul 2015TL;DR: Applied to a state-of-the-art image classification model, Batch Normalization achieves the same accuracy with 14 times fewer training steps, and beats the original model by a significant margin.
Abstract: Training Deep Neural Networks is complicated by the fact that the distribution of each layer's inputs changes during training, as the parameters of the previous layers change. This slows down the training by requiring lower learning rates and careful parameter initialization, and makes it notoriously hard to train models with saturating nonlinearities. We refer to this phenomenon as internal covariate shift, and address the problem by normalizing layer inputs. Our method draws its strength from making normalization a part of the model architecture and performing the normalization for each training mini-batch. Batch Normalization allows us to use much higher learning rates and be less careful about initialization, and in some cases eliminates the need for Dropout. Applied to a state-of-the-art image classification model, Batch Normalization achieves the same accuracy with 14 times fewer training steps, and beats the original model by a significant margin. Using an ensemble of batch-normalized networks, we improve upon the best published result on ImageNet classification: reaching 4.82% top-5 test error, exceeding the accuracy of human raters.
30,843 citations
28 Oct 2017
TL;DR: An automatic differentiation module of PyTorch is described — a library designed to enable rapid research on machine learning models that focuses on differentiation of purely imperative programs, with a focus on extensibility and low overhead.
Abstract: In this article, we describe an automatic differentiation module of PyTorch — a library designed to enable rapid research on machine learning models. It builds upon a few projects, most notably Lua Torch, Chainer, and HIPS Autograd [4], and provides a high performance environment with easy access to automatic differentiation of models executed on different devices (CPU and GPU). To make prototyping easier, PyTorch does not follow the symbolic approach used in many other deep learning frameworks, but focuses on differentiation of purely imperative programs, with a focus on extensibility and low overhead. Note that this preprint is a draft of certain sections from an upcoming paper covering all PyTorch features.
13,268 citations
Posted Content•
TL;DR: The TensorFlow interface and an implementation of that interface that is built at Google are described, which has been used for conducting research and for deploying machine learning systems into production across more than a dozen areas of computer science and other fields.
Abstract: TensorFlow is an interface for expressing machine learning algorithms, and an
implementation for executing such algorithms. A computation expressed using
TensorFlow can be executed with little or no change on a wide variety of
heterogeneous systems, ranging from mobile devices such as phones and tablets
up to large-scale distributed systems of hundreds of machines and thousands of
computational devices such as GPU cards. The system is flexible and can be used
to express a wide variety of algorithms, including training and inference
algorithms for deep neural network models, and it has been used for conducting
research and for deploying machine learning systems into production across more
than a dozen areas of computer science and other fields, including speech
recognition, computer vision, robotics, information retrieval, natural language
processing, geographic information extraction, and computational drug
discovery. This paper describes the TensorFlow interface and an implementation
of that interface that we have built at Google. The TensorFlow API and a
reference implementation were released as an open-source package under the
Apache 2.0 license in November, 2015 and are available at www.tensorflow.org.
10,447 citations