Methods › General › Loss Functions › ArcFace

Additive Angular Margin Loss

ArcFace

94 papers tagged archive 2025-07-28

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.

PaperSourceSee Code · ronghuaiyang/arcface-pytorch

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.

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.

TaskPapers
Face Recognition51
Face Verification14
Retrieval8
Image Retrieval6
Representation Learning6
Transfer Learning6
Attribute5
Face Identification5
TAR5
Triplet5
Adversarial Attack4
Data Augmentation4
Landmark Recognition4
Diversity3
Image Classification3
Image Generation3
Metric Learning3
Clustering2
Domain Adaptation2
Emotion Recognition2

Usage over time archive 2025-07-28

Papers per year tagged with ArcFace: 2018 to 2025, peak 23 23 0 2018: 1 paper 2018 2019: 8 papers 2019 2020: 14 papers 2020 2021: 19 papers 2021 2022: 14 papers 2022 2023: 23 papers 2023 2024: 10 papers 2024 2025: 5 papers 2025
Papers per year the archive tags with this method, by the paper's archive date (94 dated). Bars are counts, not a trend claim.

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

Loss Functions

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