Methods › General › Loss Functions › ArcFace
Additive Angular Margin Loss
ArcFace
Introduced by Jiankang Deng et al. in ArcFace: Additive Angular Margin Loss for Deep Face Recognition
archive 2025-07-28 Description, source and code snippet are the archive's method entry.
ArcFace, or Additive Angular Margin Loss, is a loss function used in face recognition tasks. The softmax is traditionally used in these tasks. However, the softmax loss function does not explicitly optimise the feature embedding to enforce higher similarity for intraclass samples and diversity for inter-class samples, which results in a performance gap for deep face recognition under large intra-class appearance variations.
The ArcFace loss transforms the logits Wᵀⱼxᵢ = || Wⱼ || || xᵢ || cosθⱼ, where θⱼ is the angle between the weight Wⱼ and the feature xᵢ. The individual weight || Wⱼ || = 1 is fixed by l₂ normalization. The embedding feature ||xᵢ || is fixed by l₂ normalization and re-scaled to s. The normalisation step on features and weights makes the predictions only depend on the angle between the feature and the weight. The learned embedding features are thus distributed on a hypersphere with a radius of s. Finally, an additive angular margin penalty m is added between xᵢ and W_(yᵢ) to simultaneously enhance the intra-class compactness and inter-class discrepancy. Since the proposed additive angular margin penalty is equal to the geodesic distance margin penalty in the normalised hypersphere, the method is named ArcFace:
L₃ = -1/N∑ᴺᵢ₌₁log(e^(s(cos(θ_(yᵢ) + m))))/(e^(s(cos(θ_(yᵢ) + m))) + ∑ⁿ_(j=1, j ≠yᵢ)e^(scosθⱼ))
The authors select face images from 8 different identities containing enough samples (around 1,500 images/class) to train 2-D feature embedding networks with the softmax and ArcFace loss, respectively. As the Figure shows, the softmax loss provides roughly separable feature embedding but produces noticeable ambiguity in decision boundaries, while the proposed ArcFace loss can obviously enforce a more evident gap between the nearest classes.
Other alternatives to enforce intra-class compactness and inter-class distance include Supervised Contrastive Learning.
In Syntology harvested code stored, below machine verification (unverified_code; harvested), served level V1; retrieve with get_harvested_code_for_paper(arxiv_id='1801.07698'). Repositories: 100 listed of 100 in the graph, strongest link: official. The lookup matched a paper by its title, arXiv:1801.07698, not a method entry. This answer is about arXiv:1801.07698, the paper the archive credits with this method.
Origin paper of this answer, arXiv:1801.07698 (its page on this site · its page on Syntology): Syntology's run record for its harvested code, 16 ran (of which 0 constructed an object rather than computing a result; 2 with no instrument failure: 2 honoured, 0 violated, 0 with no contract checked; 14 where Syntology's instrument failed) · 5 unverified (of 21 harvested samples) · 14 pointer-only (licence). The instrument figure counts failures of Syntology's instrument, not of the code. Ran means executed on a synthesized input, not that the code is correct.
Syntology's have check for this name (free, it answers coverage only), read 2026-10-05; ask it again live. A level is the served verification level of that one answer; nothing here is summed across methods.
What each level means, quoted from Syntology's server card: V0 “imports”; V1 “executes against a synthesized fixture and returns a value of the recorded type”; V2 “two implementations GENERATED from the paper's signature and spec agree numerically (1e-6) on fixture inputs” “neither of the two ever read the authors' own code, so V2 measures how unambiguously the paper specifies the method, not fidelity to the authors' implementation (measured 2026-09-11, commit 75eaa77: cross_checked_by is a second LLM on 2,628 of 2,831 served samples)”; V3 “paper-derived property tests pass on both of those same two generations.” The card adds: “No rung on this ladder compares the served code against the authors' implementation, so treat every level as evidence about runnability and internal agreement, not about matching the paper's results.” A level quoted in two parts is one definition the card writes as two clauses joined by a double hyphen, each clause quoted on its own (how these levels are quoted).
Papers archive 2025-07-28
archive papers tagged: 94 · with a code link: 40 · where Syntology ran a sample: 9 (7 with a run with no instrument failure, 2 where every run was a failure of Syntology's instrument) Syntology
Show: all tagged papersonly where code ran (9 of 94 tagged: 7 with a run with no instrument failure, 2 where every run was a failure of Syntology's instrument)
Page 1 of 1: papers 1 to 94 of 94, newest first by the archive's date (ties by slug), in archive order.
Repository counts are the archive's code-links table. A Syntology line states what Syntology ran from that paper's harvested code, as “N ran (of which C constructed an object rather than computing a result; K with no instrument failure: H honoured, V violated, P with no contract checked; I where Syntology's instrument failed) · U unverified”; the instrument figure counts failures of Syntology's instrument, not of the code. It is per sample and not a correctness claim. When the archive marks a repository official for the paper, the line starts with that repository's state (the archive's flag, not a verdict on who wrote the code; “community repositories only” when every sample that ran came from a community repository, “official: no sample here; runs from other or unrecorded repositories” when some came from a repository the paper names or has in its text, or from none recorded); hover it for the repositories the samples that ran came from.
-
Enhancing Few-shot Keyword Spotting Performance through Pre-Trained Self-supervised Speech Models 21 Jun 2025 · 0 repositories · arXiv:2506.17686
-
Towards Large-Scale Pose-Invariant Face Recognition Using Face Defrontalization 4 Jun 2025 · 0 repositories · arXiv:2506.04496
-
Accuracy and Fairness of Facial Recognition Technology in Low-Quality Police Images: An Experiment With Synthetic Faces 20 May 2025 · 0 repositories · arXiv:2505.14320
-
LaPIG: Cross-Modal Generation of Paired Thermal and Visible Facial Images 20 Mar 2025 · 0 repositories · arXiv:2503.16376
-
Universal Embedding Function for Traffic Classification via QUIC Domain Recognition Pretraining: A Transfer Learning Success 18 Feb 2025 · 1 repository · arXiv:2502.12930
-
Omni-ID: Holistic Identity Representation Designed for Generative Tasks 12 Dec 2024 · 0 repositories · arXiv:2412.09694
-
Multispecies Animal Re-ID Using a Large Community-Curated Dataset 7 Dec 2024 · 1 repository · arXiv:2412.05602
-
Pairwise Discernment of AffectNet Expressions with ArcFace 1 Dec 2024 · 0 repositories · arXiv:2412.01860
-
Hypersphere Secure Sketch Revisited: Probabilistic Linear Regression Attack on IronMask in Multiple Usage 19 Sep 2024 · 0 repositories · arXiv:2409.12884
-
HyperSpaceX: Radial and Angular Exploration of HyperSpherical Dimensions 5 Aug 2024 · 0 repositories · arXiv:2408.02494
-
Analyzing the Feature Extractor Networks for Face Image Synthesis 4 Jun 2024 · 1 repository · arXiv:2406.02153
-
Deep Privacy Funnel Model: From a Discriminative to a Generative Approach with an Application to Face Recognition 3 Apr 2024 · 0 repositories · arXiv:2404.02696
-
Arc2Face: A Foundation Model for ID-Consistent Human Faces 18 Mar 2024 · 3 repositories · arXiv:2403.11641Syntology community repositories only · 1 ran (of which 0 constructed an object rather than computing a result; 1 with no instrument failure: 0 honoured, 0 violated, 1 with no contract checked; 0 where Syntology's instrument failed) · 3 unverified (of 4 harvested samples) · 3 pointer-only (licence)
-
VIGFace: Virtual Identity Generation for Privacy-Free Face Recognition 13 Mar 2024 · 1 repository · arXiv:2403.08277
-
Mitigating the Impact of Attribute Editing on Face Recognition 12 Mar 2024 · 0 repositories · arXiv:2403.08092
-
X2-Softmax: Margin Adaptive Loss Function for Face Recognition 8 Dec 2023 · 0 repositories · arXiv:2312.05281
-
Improved Face Representation via Joint Label Classification and Supervised Contrastive Clustering 7 Dec 2023 · 0 repositories · arXiv:2312.04029
-
A Universal Anti-Spoofing Approach for Contactless Fingerprint Biometric Systems 23 Oct 2023 · 0 repositories · arXiv:2310.15044
-
An Empirical Study of Self-supervised Learning with Wasserstein Distance 16 Oct 2023 · 0 repositories · arXiv:2310.10143
-
Trading-off Mutual Information on Feature Aggregation for Face Recognition 22 Sep 2023 · 0 repositories · arXiv:2309.13137
-
UniTSFace: Unified Threshold Integrated Sample-to-Sample Loss for Face Recognition 21 Sep 2023 · 1 repository
-
Patent image retrieval using transformer-based deep metric learning 1 Sep 2023 · 1 repository
-
Global Features are All You Need for Image Retrieval and Reranking 14 Aug 2023 · 2 repositories · arXiv:2308.06954Syntology official (archive's flag): 14 ran · 14 ran (of which 9 constructed an object rather than computing a result; 12 with no instrument failure: 0 honoured, 0 violated, 12 with no contract checked; 2 where Syntology's instrument failed) · 4 unverified (of 18 harvested samples)
-
Improving Face Recognition from Caption Supervision with Multi-Granular Contextual Feature Aggregation 13 Aug 2023 · 0 repositories · arXiv:2308.06866
-
On the Biometric Capacity of Generative Face Models 3 Aug 2023 · 0 repositories · arXiv:2308.02065
-
PromptStyler: Prompt-driven Style Generation for Source-free Domain Generalization 27 Jul 2023 · 1 repository · arXiv:2307.15199Syntology 1 ran (of which 0 constructed an object rather than computing a result; 0 with no instrument failure: 0 honoured, 0 violated, 0 with no contract checked; 1 where Syntology's instrument failed) · 0 unverified (of 1 harvested sample)
-
RPG-Palm: Realistic Pseudo-data Generation for Palmprint Recognition 26 Jul 2023 · 0 repositories · arXiv:2307.14016
-
ChildGAN: Large Scale Synthetic Child Facial Data Using Domain Adaptation in StyleGAN 25 Jul 2023 · 0 repositories · arXiv:2307.13746
-
Imperceptible Physical Attack against Face Recognition Systems via LED Illumination Modulation 25 Jul 2023 · 0 repositories · arXiv:2307.13294
-
BlendFace: Re-designing Identity Encoders for Face-Swapping 20 Jul 2023 · 2 repositories · arXiv:2307.10854Syntology official (archive's flag): 5 ran · 5 ran (of which 3 constructed an object rather than computing a result; 5 with no instrument failure: 0 honoured, 0 violated, 5 with no contract checked; 0 where Syntology's instrument failed) · 4 unverified (of 9 harvested samples) · 9 pointer-only (licence)
-
LightFormer: An End-to-End Model for Intersection Right-of-Way Recognition Using Traffic Light Signals and an Attention Mechanism 14 Jul 2023 · 1 repository · arXiv:2307.07196
-
Towards Selection of Text-to-speech Data to Augment ASR Training 30 May 2023 · 0 repositories · arXiv:2306.00998
-
Towards minimizing efforts for Morphing Attacks -- Deep embeddings for morphing pair selection and improved Morphing Attack Detection 29 May 2023 · 0 repositories · arXiv:2305.18216
-
Human-Machine Comparison for Cross-Race Face Verification: Race Bias at the Upper Limits of Performance? 25 May 2023 · 0 repositories · arXiv:2305.16443
-
GhostFaceNets: Lightweight Face Recognition Model From Cheap Operations 10 Apr 2023 · 7 repositories
-
Patent Image Retrieval Using Cross-entropy-based Metric Learning 20 Feb 2023 · 1 repository
-
Enhancing Face Recognition with Latent Space Data Augmentation and Facial Posture Reconstruction 27 Jan 2023 · 0 repositories · arXiv:2301.11986
-
UniFace: Unified Cross-Entropy Loss for Deep Face Recognition 1 Jan 2023 · 1 repository
-
FaceQAN: Face Image Quality Assessment Through Adversarial Noise Exploration 5 Dec 2022 · 1 repository · arXiv:2212.02127
-
Supervised Fine-tuning Evaluation for Long-term Visual Place Recognition 14 Nov 2022 · 0 repositories · arXiv:2211.07696
-
Improving Deep Facial Phenotyping for Ultra-rare Disorder Verification Using Model Ensembles 12 Nov 2022 · 1 repository · arXiv:2211.06764
-
5th Place Solution to Kaggle Google Universal Image Embedding Competition 18 Oct 2022 · 1 repository · arXiv:2210.09495
-
6th Place Solution to Google Universal Image Embedding 17 Oct 2022 · 0 repositories · arXiv:2210.09377
-
3rd Place Solution for Google Universal Image Embedding 14 Oct 2022 · 0 repositories · arXiv:2210.09296
-
Can GAN-induced Attribute Manipulations Impact Face Recognition? 7 Sep 2022 · 0 repositories · arXiv:2209.02941
-
DuetFace: Collaborative Privacy-Preserving Face Recognition via Channel Splitting in the Frequency Domain 15 Jul 2022 · 1 repository · arXiv:2207.07340
-
Continuous Facial Motion Deblurring 14 Jul 2022 · 1 repository · arXiv:2207.06626
-
Are GAN-based Morphs Threatening Face Recognition? 5 May 2022 · 1 repository · arXiv:2205.02496
-
OTFace: Hard Samples Guided Optimal Transport Loss for Deep Face Representation 28 Mar 2022 · 0 repositories · arXiv:2203.14461
-
Geometric Synthesis: A Free lunch for Large-scale Palmprint Recognition Model Pretraining 11 Mar 2022 · 0 repositories · arXiv:2203.05703
-
Probabilistic Embeddings Revisited 14 Feb 2022 · 1 repository · arXiv:2202.06768Syntology official (archive's flag): 8 ran · 8 ran (of which 0 constructed an object rather than computing a result; 8 with no instrument failure: 0 honoured, 0 violated, 8 with no contract checked; 0 where Syntology's instrument failed) · 10 unverified (of 18 harvested samples)
-
Fun Selfie Filters in Face Recognition: Impact Assessment and Removal 12 Feb 2022 · 0 repositories · arXiv:2202.06022
-
ArcFace Knows the Gender, Too! 19 Dec 2021 · 0 repositories · arXiv:2112.10101
-
Distill and De-bias: Mitigating Bias in Face Verification using Knowledge Distillation 17 Dec 2021 · 0 repositories · arXiv:2112.09786
-
Producing augmentation-invariant embeddings from real-life imagery 6 Dec 2021 · 1 repository · arXiv:2112.03415Syntology official (archive's flag): 3 ran · 3 ran (of which 0 constructed an object rather than computing a result; 3 with no instrument failure: 0 honoured, 0 violated, 3 with no contract checked; 0 where Syntology's instrument failed) · 1 unverified (of 4 harvested samples)
-
Longitudinal Analysis of Mask and No-Mask on Child Face Recognition 29 Oct 2021 · 0 repositories · arXiv:2111.00121
-
A Deep Insight into Measuring Face Image Utility with General and Face-specific Image Quality Metrics 21 Oct 2021 · 0 repositories · arXiv:2110.11111
-
Google Landmark Retrieval 2021 Competition Third Place Solution 9 Oct 2021 · 0 repositories · arXiv:2110.04619
-
Efficient large-scale image retrieval with deep feature orthogonality and Hybrid-Swin-Transformers 7 Oct 2021 · 1 repository · arXiv:2110.03786
-
3rd Place Solution to Google Landmark Recognition Competition 2021 6 Oct 2021 · 0 repositories · arXiv:2110.02794
-
Dodging Attack Using Carefully Crafted Natural Makeup 14 Sep 2021 · 0 repositories · arXiv:2109.06467
-
OSSR-PID: One-Shot Symbol Recognition in P&ID Sheets using Path Sampling and GCN 8 Sep 2021 · 0 repositories · arXiv:2109.03849
-
Adversarial Relighting Against Face Recognition 18 Aug 2021 · 0 repositories · arXiv:2108.07920
-
Demographic Fairness in Face Identification: The Watchlist Imbalance Effect 15 Jun 2021 · 0 repositories · arXiv:2106.08049
-
Boosting Masked Face Recognition with Multi-Task ArcFace 20 Apr 2021 · 0 repositories · arXiv:2104.09874
-
FACESEC: A Fine-grained Robustness Evaluation Framework for Face Recognition Systems 8 Apr 2021 · 1 repository · arXiv:2104.04107Syntology official (archive's flag): 1 ran · 1 ran (of which 0 constructed an object rather than computing a result; 0 with no instrument failure: 0 honoured, 0 violated, 0 with no contract checked; 1 where Syntology's instrument failed) · 0 unverified (of 1 harvested sample) · 1 pointer-only (licence)
-
IronMask: Modular Architecture for Protecting Deep Face Template 6 Apr 2021 · 0 repositories · arXiv:2104.02239
-
Harnessing Geometric Constraints from Emotion Labels to improve Face Verification 5 Mar 2021 · 0 repositories · arXiv:2103.03862
-
When Face Recognition Meets Occlusion: A New Benchmark 4 Mar 2021 · 1 repository · arXiv:2103.02805
-
MultiFace: A Generic Training Mechanism for Boosting Face Recognition Performance 25 Jan 2021 · 1 repository · arXiv:2101.09899
-
Switchable K-Class Hyperplanes for Noise-Robust Representation Learning 1 Jan 2021 · 0 repositories
-
This Face Does Not Exist ... But It Might Be Yours! Identity Leakage in Generative Models 10 Dec 2020 · 1 repository · arXiv:2101.05084
-
Vulnerability Analysis of Face Morphing Attacks from Landmarks and Generative Adversarial Networks 9 Dec 2020 · 1 repository · arXiv:2012.05344
-
Practical No-box Adversarial Attacks against DNNs 4 Dec 2020 · 2 repositories · arXiv:2012.02525Syntology official: harvested, nothing ran · 0 ran · 2 unverified (of 2 harvested samples) · 2 pointer-only (licence)
-
FaceGuard: A Self-Supervised Defense Against Adversarial Face Images 28 Nov 2020 · 0 repositories · arXiv:2011.14218
-
Robust Attacks on Deep Learning Face Recognition in the Physical World 27 Nov 2020 · 0 repositories · arXiv:2011.13526
-
Google Landmark Recognition 2020 Competition Third Place Solution 11 Oct 2020 · 2 repositories · arXiv:2010.05350
-
Partial FC: Training 10 Million Identities on a Single Machine 11 Oct 2020 · 7 repositories · arXiv:2010.05222
-
Self-attention aggregation network for video face representation and recognition 11 Oct 2020 · 1 repository · arXiv:2010.05340
-
A2-LINK: Recognizing Disguised Faces via Active Learning and Adversarial Noise based Inter-Domain Knowledge 1 Jun 2020 · 1 repository
-
Global-Local GCN: Large-Scale Label Noise Cleansing for Face Recognition 1 Jun 2020 · 0 repositories
-
Exploring Racial Bias within Face Recognition via per-subject Adversarially-Enabled Data Augmentation 19 Apr 2020 · 0 repositories · arXiv:2004.08945
-
Google Landmarks Dataset v2 -- A Large-Scale Benchmark for Instance-Level Recognition and Retrieval 3 Apr 2020 · 5 repositories · arXiv:2004.01804
-
Improving Face Recognition from Hard Samples via Distribution Distillation Loss 10 Feb 2020 · 2 repositories · arXiv:2002.03662
-
Investigating the Impact of Inclusion in Face Recognition Training Data on Individual Face Identification 9 Jan 2020 · 0 repositories · arXiv:2001.03071
-
On adversarial patches: real-world attack on ArcFace-100 face recognition system 15 Oct 2019 · 0 repositories · arXiv:1910.07067
-
Expression, Affect, Action Unit Recognition: Aff-Wild2, Multi-Task Learning and ArcFace 25 Sep 2019 · 0 repositories · arXiv:1910.04855
-
AdvHat: Real-world adversarial attack on ArcFace Face ID system 23 Aug 2019 · 4 repositories · arXiv:1908.08705Syntology official (archive's flag): 1 ran · 4 ran (of which 0 constructed an object rather than computing a result; 1 with no instrument failure: 1 honoured, 0 violated, 0 with no contract checked; 3 where Syntology's instrument failed) · 1 unverified (of 5 harvested samples) · 3 pointer-only (licence)
-
Hard-Mining Loss based Convolutional Neural Network for Face Recognition 9 Aug 2019 · 0 repositories · arXiv:1908.09747
-
Softmax Dissection: Towards Understanding Intra- and Inter-class Objective for Embedding Learning 4 Aug 2019 · 0 repositories · arXiv:1908.01281
-
AirFace: Lightweight and Efficient Model for Face Recognition 29 Jul 2019 · 1 repository · arXiv:1907.12256
-
A New Loss Function for CNN Classifier Based on Pre-defined Evenly-Distributed Class Centroids 12 Apr 2019 · 1 repository · arXiv:1904.06008
-
A Performance Comparison of Loss Functions for Deep Face Recognition 1 Jan 2019 · 1 repository · arXiv:1901.05903
-
ArcFace: Additive Angular Margin Loss for Deep Face Recognition 23 Jan 2018 · 100 repositories · arXiv:1801.07698Syntology official: no sample here; runs from other or unrecorded repositories · 16 ran (of which 0 constructed an object rather than computing a result; 2 with no instrument failure: 2 honoured, 0 violated, 0 with no contract checked; 14 where Syntology's instrument failed) · 5 unverified (of 21 harvested samples) · 14 pointer-only (licence)
Tasks archive 2025-07-28
20 shown of 100 tasks the archive attaches to papers tagged with this method, by distinct papers. A task without a page in the catalog is plain text.
Usage over time archive 2025-07-28
Components: the archive holds no method-to-method composition, so PwC's Components table cannot be rebuilt; the Papers list carries no Results column for the same reason (the archive does not join its leaderboard rows to method tags).
Categories archive 2025-07-28
Report a problem or propose a change · a person checks every report against the paper or source before anything changes; decisions are listed on /corrections