{"url":"/sota/face-verification-on-ijb-c","task":{"name":"Face Verification","url":"/task/face-verification","note":null},"dataset":{"name":"IJB-C","url":"/dataset/ijb-c"},"category":"Computer Vision","categories":["Computer Vision"],"category_note":null,"description":"**Face Verification** is a machine learning task in computer vision that involves determining whether two facial images belong to the same person or not. The task involves extracting features from the facial images, such as the shape and texture of the face, and then using these features to compare and verify the similarity between the images.\r\n\r\n<span style=\"color:grey; opacity: 0.6\">( Image credit: [Pose-Robust Face Recognition via Deep Residual Equivariant Mapping](https://arxiv.org/pdf/1803.00839v1.pdf) )</span>","description_from":"task","source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","rank":"the archive's row order at snapshot; not re-ranked","rows_end_at":"2025-07-28","rows_withheld_as_spam":0,"metric_values":"the archive's strings, untouched"},"metrics":["TAR @ FAR=1e-6","TAR @ FAR=1e-5","TAR @ FAR=1e-4","TAR @ FAR=1e-3","TAR @ FAR=1e-2","training dataset","model","Rank-1","Rank-5"],"metric_direction":{"note":"inferred from the metric name only (the archive records no direction); null = not inferred, chart draws points only","by_metric":{"TAR @ FAR=1e-6":null,"TAR @ FAR=1e-5":null,"TAR @ FAR=1e-4":null,"TAR @ FAR=1e-3":null,"TAR @ FAR=1e-2":null,"training dataset":null,"model":null,"Rank-1":null,"Rank-5":null}},"counts":{"rows":26,"rows_with_code":24,"rows_with_paper_page":26,"rows_dated":26,"rows_using_additional_data":0},"rows":[{"rank_in_archive_order":1,"model":"HeadSharing: SH-KD","metrics":{"TAR @ FAR=1e-4":"95.64%","TAR @ FAR=1e-5":"93.73%","TAR @ FAR=1e-6":"90.24%","model":"MobileFaceNet","training dataset":"MS1M V3"},"uses_additional_data":false,"paper_date":"2022-01-18","paper":"/paper/it-s-all-in-the-head-representation-knowledge","paper_url":"https://arxiv.org/abs/2201.06945v2","paper_title":"It's All in the Head: Representation Knowledge Distillation through Classifier Sharing","code":"https://github.com/alibaba-miil/headsharingkd","n_code_links":1,"syntology":null},{"rank_in_archive_order":2,"model":"HeadSharing: TH-KD","metrics":{"TAR @ FAR=1e-4":"95.48%","TAR @ FAR=1e-5":"93.50%","TAR @ FAR=1e-6":"89.82%","model":"MobileFaceNet","training dataset":"MS1M V3"},"uses_additional_data":false,"paper_date":"2022-01-18","paper":"/paper/it-s-all-in-the-head-representation-knowledge","paper_url":"https://arxiv.org/abs/2201.06945v2","paper_title":"It's All in the Head: Representation Knowledge Distillation through Classifier Sharing","code":"https://github.com/alibaba-miil/headsharingkd","n_code_links":1,"syntology":null},{"rank_in_archive_order":3,"model":"ArcFace+CSFM","metrics":{"Rank-1":"96.31","Rank-5":"97.48","TAR @ FAR=1e-4":"95.9%","TAR @ FAR=1e-5":"94.06%","TAR @ FAR=1e-6":"89.34%"},"uses_additional_data":false,"paper_date":"2022-07-20","paper":"/paper/controllable-and-guided-face-synthesis-for","paper_url":"https://arxiv.org/abs/2207.10180v1","paper_title":"Controllable and Guided Face Synthesis for Unconstrained Face Recognition","code":"https://github.com/Faceplugin-ltd/FaceRecognition-Android","n_code_links":3,"syntology":null},{"rank_in_archive_order":4,"model":"Partial FC","metrics":{"TAR @ FAR=1e-4":"98.00%","TAR @ FAR=1e-5":"97.23%","model":"ViT-L","training dataset":"WebFace42M"},"uses_additional_data":false,"paper_date":"2022-03-28","paper":"/paper/killing-two-birds-with-one-stone-efficient","paper_url":"https://arxiv.org/abs/2203.15565v1","paper_title":"Killing Two Birds with One Stone:Efficient and Robust Training of Face Recognition CNNs by Partial FC","code":"https://github.com/deepinsight/insightface","n_code_links":6,"syntology":{"n_ran":4,"n_unverified":8,"n_samples":12,"n_pointer_only_licence":4}},{"rank_in_archive_order":5,"model":"PartialFC","metrics":{"TAR @ FAR=1e-4":"97.97%","TAR @ FAR=1e-5":"96.93%","model":"R200","training dataset":"WebFace42M"},"uses_additional_data":false,"paper_date":"2022-03-28","paper":"/paper/killing-two-birds-with-one-stone-efficient","paper_url":"https://arxiv.org/abs/2203.15565v1","paper_title":"Killing Two Birds with One Stone:Efficient and Robust Training of Face Recognition CNNs by Partial FC","code":"https://github.com/deepinsight/insightface","n_code_links":6,"syntology":{"n_ran":4,"n_unverified":8,"n_samples":12,"n_pointer_only_licence":4}},{"rank_in_archive_order":6,"model":"ArcFace","metrics":{"TAR @ FAR=1e-5":"96.07%","model":"R100","training dataset":"IBUG-500K"},"uses_additional_data":false,"paper_date":"2018-01-23","paper":"/paper/arcface-additive-angular-margin-loss-for-deep","paper_url":"https://arxiv.org/abs/1801.07698v4","paper_title":"ArcFace: Additive Angular Margin Loss for Deep Face Recognition","code":"https://github.com/deepinsight/insightface","n_code_links":100,"syntology":{"n_ran":16,"n_unverified":5,"n_samples":21,"n_pointer_only_licence":14}},{"rank_in_archive_order":7,"model":"Mag+UNPG","metrics":{"TAR @ FAR=1e-5":"94.7%"},"uses_additional_data":false,"paper_date":"2022-03-22","paper":"/paper/unified-negative-pair-generation-toward-well","paper_url":"https://arxiv.org/abs/2203.11593v2","paper_title":"Unified Negative Pair Generation toward Well-discriminative Feature Space for Face Recognition","code":"https://github.com/tomas-gajarsky/facetorch","n_code_links":2,"syntology":{"n_ran":2,"n_unverified":12,"n_samples":14,"n_pointer_only_licence":0}},{"rank_in_archive_order":8,"model":"Cos+UNPG","metrics":{"TAR @ FAR=1e-3":"97.57","TAR @ FAR=1e-4":"96.38%","TAR @ FAR=1e-5":"94.47%","model":"R100","training dataset":"MS1MV2"},"uses_additional_data":false,"paper_date":"2022-03-22","paper":"/paper/unified-negative-pair-generation-toward-well","paper_url":"https://arxiv.org/abs/2203.11593v2","paper_title":"Unified Negative Pair Generation toward Well-discriminative Feature Space for Face Recognition","code":"https://github.com/tomas-gajarsky/facetorch","n_code_links":2,"syntology":{"n_ran":2,"n_unverified":12,"n_samples":14,"n_pointer_only_licence":0}},{"rank_in_archive_order":9,"model":"L2E+IS-sampling","metrics":{"TAR @ FAR=1e-3":"97.05%","TAR @ FAR=1e-4":"95.49%","TAR @ FAR=1e-5":"93.25%","model":"MobileFaceNet","training dataset":"MS1M V3"},"uses_additional_data":false,"paper_date":"2021-10-11","paper":"/paper/rectifying-the-data-bias-in-knowledge","paper_url":"https://ieeexplore.ieee.org/document/9607686","paper_title":"Rectifying the Data Bias in Knowledge Distillation","code":null,"n_code_links":0,"syntology":null},{"rank_in_archive_order":10,"model":"MagFace++","metrics":{"TAR @ FAR=1e-4":"95.97%","TAR @ FAR=1e-5":"90.36%","model":"R100","training dataset":"MS1MV2"},"uses_additional_data":false,"paper_date":"2021-03-11","paper":"/paper/magface-a-universal-representation-for-face","paper_url":"https://arxiv.org/abs/2103.06627v4","paper_title":"MagFace: A Universal Representation for Face Recognition and Quality Assessment","code":"https://github.com/IrvingMeng/MagFace","n_code_links":2,"syntology":{"n_ran":1,"n_unverified":1,"n_samples":2,"n_pointer_only_licence":0}},{"rank_in_archive_order":11,"model":"circle loss","metrics":{"TAR @ FAR=1e-3":"96.29%","TAR @ FAR=1e-4":"93.95%","TAR @ FAR=1e-5":"89.60%","model":"R100","training dataset":"MS1M Cleaned"},"uses_additional_data":false,"paper_date":"2020-02-25","paper":"/paper/circle-loss-a-unified-perspective-of-pair","paper_url":"https://arxiv.org/abs/2002.10857v2","paper_title":"Circle Loss: A Unified Perspective of Pair Similarity Optimization","code":"https://github.com/layumi/Person_reID_baseline_pytorch","n_code_links":16,"syntology":{"n_ran":2,"n_unverified":0,"n_samples":2,"n_pointer_only_licence":1}},{"rank_in_archive_order":12,"model":"WebFace42M baseline","metrics":{"TAR @ FAR=1e-4":"97.7%","model":"R100","training dataset":"WebFace42M"},"uses_additional_data":false,"paper_date":"2021-03-06","paper":"/paper/webface260m-a-benchmark-unveiling-the-power","paper_url":"https://arxiv.org/abs/2103.04098v1","paper_title":"WebFace260M: A Benchmark Unveiling the Power of Million-Scale Deep Face Recognition","code":null,"n_code_links":0,"syntology":null},{"rank_in_archive_order":13,"model":"AdaFace (WebFace4M)","metrics":{"TAR @ FAR=1e-4":"97.39%"},"uses_additional_data":false,"paper_date":"2022-04-03","paper":"/paper/adaface-quality-adaptive-margin-for-face","paper_url":"https://arxiv.org/abs/2204.00964v2","paper_title":"AdaFace: Quality Adaptive Margin for Face Recognition","code":"https://github.com/mk-minchul/adaface","n_code_links":9,"syntology":{"n_ran":17,"n_unverified":5,"n_samples":22,"n_pointer_only_licence":11}},{"rank_in_archive_order":14,"model":"FFC","metrics":{"TAR @ FAR=1e-4":"97.31%","model":"R100","training dataset":"WebFace42M"},"uses_additional_data":false,"paper_date":"2021-05-21","paper":"/paper/an-efficient-training-approach-for-very-large","paper_url":"https://arxiv.org/abs/2105.10375v5","paper_title":"An Efficient Training Approach for Very Large Scale Face Recognition","code":"https://github.com/tiandunx/FFC","n_code_links":1,"syntology":null},{"rank_in_archive_order":15,"model":"CAFace+AdaFace (WebFace4M)","metrics":{"TAR @ FAR=1e-3":"98.08","TAR @ FAR=1e-4":"97.3%"},"uses_additional_data":false,"paper_date":"2022-10-19","paper":"/paper/cluster-and-aggregate-face-recognition-with","paper_url":"https://arxiv.org/abs/2210.10864v3","paper_title":"Cluster and Aggregate: Face Recognition with Large Probe Set","code":"https://github.com/mk-minchul/caface","n_code_links":1,"syntology":{"n_ran":7,"n_unverified":2,"n_samples":9,"n_pointer_only_licence":0}},{"rank_in_archive_order":16,"model":"AdaFace (MS1MV3)","metrics":{"TAR @ FAR=1e-4":"97.09%"},"uses_additional_data":false,"paper_date":"2022-04-03","paper":"/paper/adaface-quality-adaptive-margin-for-face","paper_url":"https://arxiv.org/abs/2204.00964v2","paper_title":"AdaFace: Quality Adaptive Margin for Face Recognition","code":"https://github.com/mk-minchul/adaface","n_code_links":9,"syntology":{"n_ran":17,"n_unverified":5,"n_samples":22,"n_pointer_only_licence":11}},{"rank_in_archive_order":17,"model":"AdaFace (MS1MV2)","metrics":{"TAR @ FAR=1e-4":"96.89%"},"uses_additional_data":false,"paper_date":"2022-04-03","paper":"/paper/adaface-quality-adaptive-margin-for-face","paper_url":"https://arxiv.org/abs/2204.00964v2","paper_title":"AdaFace: Quality Adaptive Margin for Face Recognition","code":"https://github.com/mk-minchul/adaface","n_code_links":9,"syntology":{"n_ran":17,"n_unverified":5,"n_samples":22,"n_pointer_only_licence":11}},{"rank_in_archive_order":18,"model":"ElasticFace-Cos","metrics":{"TAR @ FAR=1e-4":"96.57%","model":"R100","training dataset":"MS1M V2"},"uses_additional_data":false,"paper_date":"2021-09-20","paper":"/paper/elasticface-elastic-margin-loss-for-deep-face","paper_url":"https://arxiv.org/abs/2109.09416v4","paper_title":"ElasticFace: Elastic Margin Loss for Deep Face Recognition","code":"https://github.com/fdbtrs/ElasticFace","n_code_links":3,"syntology":null},{"rank_in_archive_order":19,"model":"Arc+UNPG","metrics":{"TAR @ FAR=1e-3":"97.51","TAR @ FAR=1e-4":"96.33%"},"uses_additional_data":false,"paper_date":"2022-03-22","paper":"/paper/unified-negative-pair-generation-toward-well","paper_url":"https://arxiv.org/abs/2203.11593v2","paper_title":"Unified Negative Pair Generation toward Well-discriminative Feature Space for Face Recognition","code":"https://github.com/tomas-gajarsky/facetorch","n_code_links":2,"syntology":{"n_ran":2,"n_unverified":12,"n_samples":14,"n_pointer_only_licence":0}},{"rank_in_archive_order":20,"model":"QMagFace","metrics":{"TAR @ FAR=1e-2":"98.51","TAR @ FAR=1e-3":"97.62","TAR @ FAR=1e-4":"96.19%"},"uses_additional_data":false,"paper_date":"2021-11-26","paper":"/paper/qmagface-simple-and-accurate-quality-aware","paper_url":"https://arxiv.org/abs/2111.13475v3","paper_title":"QMagFace: Simple and Accurate Quality-Aware Face Recognition","code":"https://github.com/pterhoer/QMagFace","n_code_links":1,"syntology":null},{"rank_in_archive_order":21,"model":"CurricularFace","metrics":{"TAR @ FAR=1e-4":"96.1%"},"uses_additional_data":false,"paper_date":"2020-04-01","paper":"/paper/curricularface-adaptive-curriculum-learning","paper_url":"https://arxiv.org/abs/2004.00288v1","paper_title":"CurricularFace: Adaptive Curriculum Learning Loss for Deep Face Recognition","code":"https://github.com/HuangYG123/CurricularFace","n_code_links":2,"syntology":{"n_ran":5,"n_unverified":8,"n_samples":13,"n_pointer_only_licence":0}},{"rank_in_archive_order":22,"model":"PFEfuse + match","metrics":{"TAR @ FAR=1e-2":"97.17%","TAR @ FAR=1e-3":"95.49%","model":"SphereFace64","training dataset":"MS1M V2"},"uses_additional_data":false,"paper_date":"2019-04-21","paper":"/paper/probabilistic-face-embeddings","paper_url":"https://arxiv.org/abs/1904.09658v4","paper_title":"Probabilistic Face Embeddings","code":"https://github.com/seasonSH/Probabilistic-Face-Embeddings","n_code_links":1,"syntology":{"n_ran":1,"n_unverified":0,"n_samples":1,"n_pointer_only_licence":0}},{"rank_in_archive_order":23,"model":"VGGFace2_ft","metrics":{"TAR @ FAR=1e-2":"96.7%","TAR @ FAR=1e-3":"92.7%","model":"R50","training dataset":"Vggface2"},"uses_additional_data":false,"paper_date":"2017-10-23","paper":"/paper/vggface2-a-dataset-for-recognising-faces","paper_url":"http://arxiv.org/abs/1710.08092v2","paper_title":"VGGFace2: A dataset for recognising faces across pose and age","code":"https://github.com/deepinsight/insightface","n_code_links":24,"syntology":{"n_ran":5,"n_unverified":3,"n_samples":8,"n_pointer_only_licence":5}},{"rank_in_archive_order":24,"model":"AIM","metrics":{"TAR @ FAR=1e-2":"93.5%"},"uses_additional_data":false,"paper_date":"2018-09-02","paper":"/paper/look-across-elapse-disentangled","paper_url":"http://arxiv.org/abs/1809.00338v2","paper_title":"Look Across Elapse: Disentangled Representation Learning and Photorealistic Cross-Age Face Synthesis for Age-Invariant Face Recognition","code":"https://github.com/ZhaoJ9014/High_Performance_Face_Recognition","n_code_links":1,"syntology":{"n_ran":0,"n_unverified":6,"n_samples":6,"n_pointer_only_licence":0}},{"rank_in_archive_order":25,"model":"MN-vc","metrics":{"TAR @ FAR=1e-2":"92.70%"},"uses_additional_data":false,"paper_date":"2018-07-24","paper":"/paper/multicolumn-networks-for-face-recognition","paper_url":"http://arxiv.org/abs/1807.09192v1","paper_title":"Multicolumn Networks for Face Recognition","code":"https://github.com/ibendrup/MulticolumnNetwork","n_code_links":1,"syntology":{"n_ran":0,"n_unverified":4,"n_samples":4,"n_pointer_only_licence":0}},{"rank_in_archive_order":26,"model":"FaceNet","metrics":{"TAR @ FAR=1e-2":"66.5%"},"uses_additional_data":false,"paper_date":"2015-03-12","paper":"/paper/facenet-a-unified-embedding-for-face","paper_url":"http://arxiv.org/abs/1503.03832v3","paper_title":"FaceNet: A Unified Embedding for Face Recognition and Clustering","code":"https://github.com/serengil/deepface","n_code_links":183,"syntology":{"n_ran":65,"n_unverified":89,"n_samples":154,"n_pointer_only_licence":45}}],"since_archive":{"present":false,"note":"No Syntology-extracted rows are published in this build."},"syntology":{"read_at":"2026-09-24T18:15:14+00:00","claim":"Per row: N of M harvested code samples from that row's paper executed on a synthesized fixture; the other M-N are unverified. Not a reproduction of the row's number; not a correctness claim. n_pointer_only_licence counts samples the site points at rather than redistributes (a licence axis, independent of ran/unverified).","rows_with_graph_line":18,"rows_with_any_sample_ran":16,"distinct_papers_with_graph_line":13,"distinct_papers_with_any_sample_ran":11,"samples_over_distinct_papers":{"n_ran":125,"n_unverified":143,"n_samples":268,"n_pointer_only_licence":80,"note":"each paper (arXiv id) counted once, however many rows it is behind; this is the page-level figure"},"samples_row_weighted":{"n_ran":167,"n_unverified":185,"n_samples":352,"n_pointer_only_licence":106,"note":"row-weighted: a paper behind several rows is counted once per row; inflated relative to samples_over_distinct_papers by design, kept for readers summing the per-row syntology blocks"}}}