mlsanityData to 14 Oct 2026

Efficient, Validation-Free Intrinsic Quality Estimation for Large-Scale Face Recognition Datasets

arXiv 2605.29720 ↗First posted 28 May 2026cs.CV cs.LGcited by 0

Not on the map. Why?

Abstract (arXiv)

We propose Intrinsic Quality (IQ), a validation-free metric designed to estimate the inherent potential of face recognition (FR) datasets to produce high-performance models without the need for full-scale training. IQ integrates two components: (i) a Neighbor-Consistency Score that quantifies local identity label agreement via nearest neighbors, and (ii) Global Representation Subspace Complexity (Effective Rank, ER), which captures the underlying embedding geometry and dataset diversity. IQ allows for rapid evaluation using lightweight proxy models or data subsets, facilitating dataset diagnosis and curation prior to resource-intensive full-scale training. We describe an experimental protocol tailored to clean, noisy, and mixed-quality FR datasets, and outline evaluation methodologies to validate IQ's predictive power for downstream performance.

Key facts. No citing papers in the map yet (data through 14 Oct 2026). Builds on 29 of its 41 references that are in the map. First posted on arXiv 28 May 2026.

Builds on

29 of 41 references.

References · 29

  1. 2015Deep Residual Learning for Image Recognitioncited by 52,096
  2. 2018CosFace: Large Margin Cosine Loss for Deep Face Recognitioncited by 971
  3. 2017VGGFace2: A dataset for recognising faces across pose and agecited by 914
  4. 2017SphereFace: Deep Hypersphere Embedding for Face Recognitioncited by 737
  5. 2023DataComp: In search of the next generation of multimodal datasetscited by 644
  6. 2019Data Shapley: Equitable Valuation of Data for Machine Learningcited by 629
  7. 2016MS-Celeb-1M: A Dataset and Benchmark for Large-Scale Face Recognitioncited by 624
  8. 2020Dataset Cartography: Mapping and Diagnosing Datasets with Training Dynamicscited by 425
  9. 2019Confident Learning: Estimating Uncertainty in Dataset Labelscited by 413
  10. 2021Pervasive Label Errors in Test Sets Destabilize Machine Learning Benchmarkscited by 395
  11. 2021MagFace: A Universal Representation for Face Recognition and Quality Assessmentcited by 334
  12. 2021WebFace260M: A Benchmark Unveiling the Power of Million-Scale Deep Face Recognitioncited by 223
17 more
  1. 2018Deep k-Nearest Neighbors: Towards Confident, Interpretable and Robust Deep Learningcited by 207
  2. 2021General Facial Representation Learning in a Visual-Linguistic Mannercited by 195
  3. 2020LEEP: A New Measure to Evaluate Transferability of Learned Representationscited by 188
  4. 2015A Light CNN for Deep Face Representation with Noisy Labelscited by 126
  5. 2020SER-FIQ: Unsupervised Estimation of Face Image Quality Based on Stochastic Embedding Robustnesscited by 112
  6. 2020Deep k-NN for Noisy Labelscited by 55
  7. 2022Learning with Neighbor Consistency for Noisy Labelscited by 55
  8. 2021Frustratingly Easy Transferability Estimationcited by 45
  9. 2024Diff-eRank: A Novel Rank-Based Metric for Evaluating Large Language Modelscited by 35
  10. 2022How stable are Transferability Metrics evaluations?cited by 30
  11. 2023Matrix Information Theory for Self-Supervised Learningcited by 21
  12. 2023Toward High Quality Facial Representation Learningcited by 10
  13. 2023Unsupervised Embedding Quality Evaluationcited by 9
  14. 2022BoundaryFace: A mining framework with noise label self-correction for Face Recognitioncited by 6
  15. 2024An Embedding is Worth a Thousand Noisy Labelscited by 5
  16. 2025X-Factor: Quality Is a Dataset-Intrinsic Propertycited by 3
  17. 2024RepFace: Refining Closed-Set Noise with Progressive Label Correction for Face Recognitioncited by 1

Built on by

Not cited yet.