IEEE Neural Networks Council
Standards Committee

Working Group on Pattern Recognition Benchmarks



Goal

The goal of the Working Group on Pattern Recognition Benchmarks (under IEEE Neural Networks Council Standards Committee) is to provide easy access to data sets and algorithms which may be used for benchmarking and comparison purposes via the WWW.

This page is constantly under construction and the lists of datasets, algorithms (code) and publications are by no means complete. If you have new datasets/code/papers and would like to put them here, please send me an email (hall@csee.usf.edu) to let me know where to link them. Any feedbacks and suggestions are also highly welcome.

Lawrence O. Hall
CSE Dept., University of South Florida, Tampa, Fl.



General Data Sets

(In no particular order)

UCI Machine Learning Repository
DELVE Dataset
UCID is intended as a benchmark dataset for image retrieval. All images are preserved in their uncompressed form hence making the dataset also useful for evaluating image compression, colour quantisation and other algorithms.
The Finnish Neural Net Archive
CMU Neural Network - Benchmark Collection
ELENA Learning Database: FTP and README
The NN learning algorithm benchmarking page
NIST
University of Texas at Austin Neural Net Archive
UCSD Neural Net Archive
Neuron Digest
DIMUND Document Image Understanding Information Server
UNIPEN Handwriting data and software
Benchmarking texture classification algorithms.
MLnet Machine Learning Archive at GMD
Jacob, Content-based browsing and querying in video databases.
Neural Networks, Connectionist Systems, and Neural Systems from the CMU AI repository. Includes code.


Web-based algorithm evaluation



  • Web-based algorithm evaluation


    Pointers to Publications (Papers, bibliographies and Software)

    (In no particular order)




    Here are some Mammography data sets.

    Here are some Document Analysis and Pattern Recognition related WWW server from around the world

    Internet Sources for Range Data

    Follow the links below to find range images in various formats.

  • Michigan State University archive.
  • PEIPA (Essex, UK) archive (includes MSU/WSU and Edinburgh images)
  • University of Utah Range Database
  • Registered range/intensity pairs from MSU
  • USC Image data archive


  • Face Recognition

    Freeware

    Internet Resources: