> ML_DATASET // CIFAR-100_v1.0
CIFAR-100 Dataset (Krizhevsky 2009)
University of Toronto (Alex Krizhevsky) · Computer Vision Education · 60,000 32x32 color images across 100 classes grouped into 20 superclasses
Computer Vision EducationMIT / Open Academic60,000 32x32 color images across 100 classes grouped into 20 superclassesopen
Dataset Profile & Characteristics
Label Type:Fine class label (100) and coarse superclass label (20)
Languages:en
License Tier:permissive-open-source
Modalities:image
Intended Use
- Fine-grained classification research and few-shot representation learning benchmarks
Prohibited / Discouraged Use
- Production vision systems requiring high-resolution textures
Bias, Leakage & Privacy Risk Analysis
Privacy / Sensitive Data Risks:
Tiny 32x32 resolution images; zero PII.
Known Bias:
Low 32x32 resolution limitations; 500 training images per fine class.
Known Benchmark Leakage:
Standardized train/test split.