> ML_DATASET // CELEBA-DATASET_v1.0
CelebFaces Attributes Dataset (CelebA)
Chinese University of Hong Kong (Liu et al.) · Face Recognition & Attribute Prediction · 202,599 face images of 10,177 celebrity identities with 40 binary attributes
Face Recognition & Attribute PredictionNon-Commercial Research License202,599 face images of 10,177 celebrity identities with 40 binary attributesopen
Dataset Profile & Characteristics
Label Type:40 binary facial attributes, 5 landmark locations, identity labels
Languages:en
License Tier:non-commercial
Modalities:image
Intended Use
- Generative model facial synthesis evaluation (GANs, VAEs, Diffusion)
- Facial landmark detection benchmarking
Prohibited / Discouraged Use
- Biometric surveillance or facial attribute profiling without consent
Bias, Leakage & Privacy Risk Analysis
Privacy / Sensitive Data Risks:
Web-scraped celebrity photos; facial recognition privacy and consent implications.
Known Bias:
Heavy skew towards young, attractive, Caucasian celebrities with professional studio lighting.
Known Benchmark Leakage:
Identity overlaps across train/val/test splits must be strictly partitioned by identity.