Call for NIPS 2008 Kernel Learning Workshop Submissions
Document Actions
Data
Datasets for Benchmarks and Applications
- DaimlerChrysler Pedestrian Classification Benchmark. Large dataset (4000 pedestrian and 5000 non-pedestrian samples of size 18x36, plus 1200 additional images) for benchmarking classification methods.
- Chairs. Chair Database of the Max-Planck-Institut for Biological Cybernetics, Tübingen.
- Delve. Data for Evaluating Learning in Valid Experiments, Toronto.
- MNIST. Optical Character Database at AT&T Research.
- NIST. Dataset of Handwritten Characters, created by the National Institute of Standards Bureau, 591 MB.
- USPS on FTP site. Dataset of Handwritten Characters, created by the US Postal Service, including C routines for loading/saving.
- Reuters-21578. Reuters 21578 Text Categorization Test Collection.
- STATLIB. Datasets from the Statistics Department at CMU.
- USPS. Dataset of Handwritten Characters, created by the US Postal Service, 29 MB.