# Face Verification on MegaFace Challenge 1 Source: [Face Verification on MegaFace Challenge 1]() Collection index: [llms.txt]() Data updated: 2026-02-26T06:05:50.314Z Evaluation of face verification performance on the MegaFace Challenge 1 dataset, typically measured using True Acceptance Rate (TAR) at a False Acceptance Rate (FAR) of 10^-6. A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage. | Scope | Recorded value | | --- | --- | | Task | Face Verification | | Dataset (including any stated split) | MegaFace Challenge 1 | | Metric | TAR (@10^-6 FAR) | | Unit | PERCENT | | Direction | Higher is better | | Results loaded | 28 of 28 | | Coverage | All results currently returned for this benchmark | Evaluation settings may differ between results. The original papers and recorded configurations are needed to establish comparability. A shared metric name alone does not establish an identical protocol. ## Metrics - [TAR (@10^-6 FAR)](): Higher is better; unit: PERCENT. ## Recorded results | Method | TAR (@10^-6 FAR) | Result date | Report paper | | --- | --- | --- | --- | | PPRN | 99.1 | Not recorded | 68fa692e1e209131cea20999 | | PPRN | 98.98 | Not recorded | 68fa692e1e209131cea20999 | | GroupFace | 98.79 | Not recorded | 68fa692e1e209131cea20999 | | CircleLoss | 98.73 | Not recorded | 68fa692e1e209131cea20999 | | CurricularFace | 98.64 | Not recorded | 68fa692e1e209131cea20999 | | PPRN | 98.59 | Not recorded | 68fa692e1e209131cea20999 | | PPRN | 98.58 | Not recorded | 68fa692e1e209131cea20999 | | PPRN | 98.49 | Not recorded | 68fa692e1e209131cea20999 | | ArcFace | 98.48 | Not recorded | 68fa692e1e209131cea20999 | | PPRN | 98.03 | Not recorded | 68fa692e1e209131cea20999 | | CosFace(3-patch ensemble) | 97.96 | Not recorded | 68fa642c1e209131cea031c6 | | CosFace | 97.91 | Not recorded | 68fa692e1e209131cea20999 | | CosFace(Single-patch) | 96.65 | Not recorded | 68fa642c1e209131cea031c6 | | DeepSense V2 | 95.99 | Not recorded | 68fa642c1e209131cea031c6 | | Vocord - deepVo V3 | 94.96 | Not recorded | 68fa642c1e209131cea031c6 | | CosFace(3-patch ensemble) | 92.22 | Not recorded | 68fa642c1e209131cea031c6 | | YouTu Lab | 91.34 | Not recorded | 68fa642c1e209131cea031c6 | | SphereFace - Small | 90.04 | Not recorded | 68fa642c1e209131cea031c6 | | CosFace(Single-patch) | 89.88 | Not recorded | 68fa642c1e209131cea031c6 | | SIATMMLAB Tencent Vision | 87.27 | Not recorded | 68fa642c1e209131cea031c6 | | Google - FaceNet v8 | 86.47 | Not recorded | 68fa642c1e209131cea031c6 | | NTechLAB - facenx\_large | 85.08 | Not recorded | 68fa642c1e209131cea031c6 | | DeepSense - Small | 82.85 | Not recorded | 68fa642c1e209131cea031c6 | | FUDAN-CS\_SDS | 79.19 | Not recorded | 68fa642c1e209131cea031c6 | | Beijing FaceAll V2 | 77.6 | Not recorded | 68fa642c1e209131cea031c6 | | SIAT\_MMLAB | 76.72 | Not recorded | 68fa642c1e209131cea031c6 | | GRCCV | 74.88 | Not recorded | 68fa642c1e209131cea031c6 | | Beijing FaceAll\_Norm\_1600 | 67.11 | Not recorded | 68fa642c1e209131cea031c6 | ### PPRN TAR (@10^-6 FAR): 99.1. Result ID: 698d9bda8b517eb43322a7cc. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Training Dataset":"Glint360K","Loss Function":"CosFace","Sampling Rate (r)":"0.1"}. This record does not assert independent reproduction. ### PPRN TAR (@10^-6 FAR): 98.98. Result ID: 698d9bda8b517eb43322a7cb. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Training Dataset":"Glint360K","Loss Function":"CosFace","Sampling Rate (r)":"1.0"}. This record does not assert independent reproduction. ### GroupFace TAR (@10^-6 FAR): 98.79. Result ID: 698d9bda8b517eb43322a7c4. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Loss Function":"GroupFace"}. This record does not assert independent reproduction. ### CircleLoss TAR (@10^-6 FAR): 98.73. Result ID: 698d9bda8b517eb43322a7c5. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Loss Function":"CircleLoss"}. This record does not assert independent reproduction. ### CurricularFace TAR (@10^-6 FAR): 98.64. Result ID: 698d9bda8b517eb43322a7c6. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Loss Function":"CurricularFace"}. This record does not assert independent reproduction. ### PPRN TAR (@10^-6 FAR): 98.59. Result ID: 698d9bda8b517eb43322a7c9. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Training Dataset":"MS1MV2","Loss Function":"ArcFace","Sampling Rate (r)":"1.0"}. This record does not assert independent reproduction. ### PPRN TAR (@10^-6 FAR): 98.58. Result ID: 698d9bda8b517eb43322a7c7. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Training Dataset":"MS1MV2","Loss Function":"CosFace","Sampling Rate (r)":"1.0"}. This record does not assert independent reproduction. ### PPRN TAR (@10^-6 FAR): 98.49. Result ID: 698d9bda8b517eb43322a7c8. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Training Dataset":"MS1MV2","Loss Function":"CosFace","Sampling Rate (r)":"0.1"}. This record does not assert independent reproduction. ### ArcFace TAR (@10^-6 FAR): 98.48. Result ID: 698d9bda8b517eb43322a7c3. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Loss Function":"ArcFace (0.5)"}. This record does not assert independent reproduction. ### PPRN TAR (@10^-6 FAR): 98.03. Result ID: 698d9bda8b517eb43322a7ca. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Training Dataset":"MS1MV2","Loss Function":"ArcFace","Sampling Rate (r)":"0.1"}. This record does not assert independent reproduction. ### CosFace(3-patch ensemble) TAR (@10^-6 FAR): 97.96. Result ID: 698c0f77f729bc7f68e81171. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Protocol":"Large"}. This record does not assert independent reproduction. ### CosFace TAR (@10^-6 FAR): 97.91. Result ID: 698d9bda8b517eb43322a7c2. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Loss Function":"CosFace (0.35)"}. This record does not assert independent reproduction. ### CosFace(Single-patch) TAR (@10^-6 FAR): 96.65. Result ID: 698c0f77f729bc7f68e81170. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Protocol":"Large"}. This record does not assert independent reproduction. ### DeepSense V2 TAR (@10^-6 FAR): 95.99. Result ID: 698c0f77f729bc7f68e8116d. [Reporting paper]() Recorded evaluation configuration: {"Protocol":"Large"}. This record does not assert independent reproduction. ### Vocord - deepVo V3 TAR (@10^-6 FAR): 94.96. Result ID: 698c0f77f729bc7f68e8116f. [Reporting paper]() Recorded evaluation configuration: {"Protocol":"Large"}. This record does not assert independent reproduction. ### CosFace(3-patch ensemble) TAR (@10^-6 FAR): 92.22. Result ID: 698c0f77f729bc7f68e81168. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Protocol":"Small"}. This record does not assert independent reproduction. ### YouTu Lab TAR (@10^-6 FAR): 91.34. Result ID: 698c0f77f729bc7f68e8116e. [Reporting paper]() Recorded evaluation configuration: {"Protocol":"Large"}. This record does not assert independent reproduction. ### SphereFace - Small TAR (@10^-6 FAR): 90.04. Result ID: 698c0f77f729bc7f68e81163. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Protocol":"Small"}. This record does not assert independent reproduction. ### CosFace(Single-patch) TAR (@10^-6 FAR): 89.88. Result ID: 698c0f77f729bc7f68e81167. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Protocol":"Small"}. This record does not assert independent reproduction. ### SIATMMLAB Tencent Vision TAR (@10^-6 FAR): 87.27. Result ID: 698c0f77f729bc7f68e8116c. [Reporting paper]() Recorded evaluation configuration: {"Protocol":"Large"}. This record does not assert independent reproduction. ### Google - FaceNet v8 TAR (@10^-6 FAR): 86.47. Result ID: 698c0f77f729bc7f68e8116a. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Protocol":"Large"}. This record does not assert independent reproduction. ### NTechLAB - facenx\_large TAR (@10^-6 FAR): 85.08. Result ID: 698c0f77f729bc7f68e8116b. [Reporting paper]() Recorded evaluation configuration: {"Protocol":"Large"}. This record does not assert independent reproduction. ### DeepSense - Small TAR (@10^-6 FAR): 82.85. Result ID: 698c0f77f729bc7f68e81162. [Reporting paper]() Recorded evaluation configuration: {"Protocol":"Small"}. This record does not assert independent reproduction. ### FUDAN-CS\_SDS TAR (@10^-6 FAR): 79.19. Result ID: 698c0f77f729bc7f68e81166. [Reporting paper]() Recorded evaluation configuration: {"Protocol":"Small"}. This record does not assert independent reproduction. ### Beijing FaceAll V2 TAR (@10^-6 FAR): 77.6. Result ID: 698c0f77f729bc7f68e81164. [Reporting paper]() Recorded evaluation configuration: {"Protocol":"Small"}. This record does not assert independent reproduction. ### SIAT\_MMLAB TAR (@10^-6 FAR): 76.72. Result ID: 698c0f77f729bc7f68e81161. [Reporting paper]() Recorded evaluation configuration: {"Protocol":"Small"}. This record does not assert independent reproduction. ### GRCCV TAR (@10^-6 FAR): 74.88. Result ID: 698c0f77f729bc7f68e81165. [Reporting paper]() Recorded evaluation configuration: {"Protocol":"Small"}. This record does not assert independent reproduction. ### Beijing FaceAll\_Norm\_1600 TAR (@10^-6 FAR): 67.11. Result ID: 698c0f77f729bc7f68e81169. [Reporting paper]() Recorded evaluation configuration: {"Protocol":"Large"}. This record does not assert independent reproduction. Page 1 of 1. This is an index page, not the full collection.