> ML_DATASET // LABELED-FACES-IN-THE-WILD_v1.0
Labeled Faces in the Wild (LFW)
University of Massachusetts, Amherst (Huang et al.) · Face Verification Benchmark · 13,233 face images of 5,749 people collected from the web
Face Verification BenchmarkOpen for Non-Commercial Research13,233 face images of 5,749 people collected from the webopen
Dataset Profile & Characteristics
Label Type:Identity labels with standardized 6,000 pair verification splits
Languages:en
License Tier:non-commercial
Modalities:image
Intended Use
- Classic academic benchmark for unconstrained facial verification (same/different identity)
Prohibited / Discouraged Use
- Production commercial biometric authentication without modern demographic parity testing
Bias, Leakage & Privacy Risk Analysis
Privacy / Sensitive Data Risks:
Identifiable public figures, politicians, and celebrities photographed in news media.
Known Bias:
Skewed toward male, middle-aged public figures; news media lighting and poses.
Known Benchmark Leakage:
Strict 10-fold cross-validation verification pairs prevent subject overlap.