mlsanityData to 7 Oct 2026

The Devil of Face Recognition is in the Noise

arXiv 1807.11649 ↗First posted 31 Jul 2018cs.CVcited by 65

Not on the map. Why?

Abstract (arXiv)

The growing scale of face recognition datasets empowers us to train strong convolutional networks for face recognition. While a variety of architectures and loss functions have been devised, we still have a limited understanding of the source and consequence of label noise inherent in existing datasets. We make the following contributions: 1) We contribute cleaned subsets of popular face databases, i.e., MegaFace and MS-Celeb-1M datasets, and build a new large-scale noise-controlled IMDb-Face dataset. 2) With the original datasets and cleaned subsets, we profile and analyze label noise properties of MegaFace and MS-Celeb-1M. We show that a few orders more samples are needed to achieve the same accuracy yielded by a clean subset. 3) We study the association between different types of noise, i.e., label flips and outliers, with the accuracy of face recognition models. 4) We investigate ways to improve data cleanliness, including a comprehensive user study on the influence of data labeling strategies to annotation accuracy. The IMDb-Face dataset has been released on https://github.com/fwang91/IMDb-Face.

Key facts. Cited by 65 papers in the map (data through 7 Oct 2026). First posted on arXiv 31 Jul 2018.

Builds on

No reference list yet.

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2021: 23; 2022: 11; 2023: 10; 2024: 9; 2025: 7; 2026: 5.

65 papers.

Sort:

By area:

  1. 2021WebFace260M: A Benchmark Unveiling the Power of Million-Scale Deep Face Recognitioncited by 221
  2. 2021SynFace: Face Recognition with Synthetic Datacited by 106
  3. 2022Killing Two Birds with One Stone:Efficient and Robust Training of Face Recognition CNNs by Partial FCcited by 73
  4. 2022WebFace260M: A Benchmark for Million-Scale Deep Face Recognitioncited by 41
  5. 2023Synthetic Data for Face Recognition: Current State and Future Prospectscited by 38
  6. 2021Pre-training strategies and datasets for facial representation learningcited by 25
  7. 2021Federated Learning for Face Recognition with Gradient Correctioncited by 24
  8. 2021Compatibility-aware Heterogeneous Visual Searchcited by 22
  9. 2022SFace: Sigmoid-Constrained Hypersphere Loss for Robust Face Recognitioncited by 21
  10. 2023Deep Learning for Camera Calibration and Beyond: A Surveycited by 20
  11. 2021Joint Face Image Restoration and Frontalization for Recognitioncited by 18
  12. 2021Learning Hierarchical Graph Neural Networks for Image Clusteringcited by 18
38 more
  1. 2023Neuromorphic Event-based Facial Expression Recognitioncited by 18
  2. 2021Cross-Domain Similarity Learning for Face Recognition in Unseen Domainscited by 13
  3. 2024The METRIC-framework for assessing data quality for trustworthy AI in medicine: a systematic reviewcited by 13
  4. 2021An Efficient Training Approach for Very Large Scale Face Recognitioncited by 12
  5. 2021Masked Face Recognition Challenge: The WebFace260M Track Reportcited by 12
  6. 2024Fairness and Bias Mitigation in Computer Vision: A Surveycited by 12
  7. 2021Exploring Robust Architectures for Deep Artificial Neural Networkscited by 11
  8. 2021EQFace: A Simple Explicit Quality Network for Face Recognitioncited by 10
  9. 2024Synthesizing Efficient Data with Diffusion Models for Person Re-Identification Pre-Trainingcited by 10
  10. 2021Towards Robustness to Label Noise in Text Classification via Noise Modelingcited by 9
  11. 2021Dual-Tuning: Joint Prototype Transfer and Structure Regularization for Compatible Feature Learningcited by 8
  12. 2025Beyond Real Faces: Synthetic Datasets Can Achieve Reliable Recognition Performance without Privacy Compromisecited by 8
  13. 2021Harnessing Unrecognizable Faces for Improving Face Recognitioncited by 7
  14. 2022BoundaryFace: A mining framework with noise label self-correction for Face Recognitioncited by 6
  15. 2023Metric Compatible Training for Online Backfilling in Large-Scale Retrievalcited by 6
  16. 2021Recent Advances in Deep Learning Techniques for Face Recognitioncited by 5
  17. 2024VIGFace: Virtual Identity Generation for Privacy-Free Face Recognitioncited by 5
  18. 2023ADD: An Automatic Desensitization Fisheye Dataset for Autonomous Drivingcited by 4
  19. 2025HiFi-Portrait: Zero-shot Identity-preserved Portrait Generation with High-fidelity Multi-face Fusioncited by 4
  20. 2021DAIL: Dataset-Aware and Invariant Learning for Face Recognitioncited by 3
  21. 2021Prototype Memory for Large-scale Face Representation Learningcited by 3
  22. 2022A comprehensive survey on semantic facial attribute editing using generative adversarial networkscited by 3
  23. 2022Median Pixel Difference Convolutional Network for Robust Face Recognitioncited by 3
  24. 2022Towards Robust Face Recognition with Comprehensive Searchcited by 3
  25. 2023Relabeling Minimal Training Subset to Flip a Predictioncited by 3
  26. 2025UIFace: Unleashing Inherent Model Capabilities to Enhance Intra-Class Diversity in Synthetic Face Recognitioncited by 3
  27. 2025Knowledge Completes the Vision: A Multimodal Entity-aware Retrieval-Augmented Generation Framework for News Image Captioningcited by 3
  28. 2021Boosting Semi-Supervised Face Recognition with Noise Robustnesscited by 2
  29. 2021MixFace: Improving Face Verification Focusing on Fine-grained Conditionscited by 2
  30. 2022Practical Digital Disguises: Leveraging Face Swaps to Protect Patient Privacycited by 2
  31. 2022Deep Image Retrieval is not Robust to Label Noisecited by 2
  32. 2023On Responsible Machine Learning Datasets with Fairness, Privacy, and Regulatory Normscited by 2
  33. 2025GIF: Generative Inspiration for Face Recognition at Scalecited by 2
  34. 2021How Unique Is a Face: An Investigative Studycited by 1
  35. 2021Evaluation of Human and Machine Face Detection using a Novel Distinctive Human Appearance Datasetcited by 1
  36. 2022A Survey of Face Recognitioncited by 1
  37. 2023Face Recognition Using Synthetic Face Datacited by 1
  38. 2023Towards Large-scale Masked Face Recognitioncited by 1

50 most cited of 65 All 65 (JSON)