Monday June 18, 2018

You may also be interested in:
The XM2VTS score-level Fusion Benchmark Database
The Biosecure DS2 score+quality Database
Return to the main page

BANCA score database

Quick summary

Following the permanent lost of the original address (, I put the reformated scores of the BANCA database online on this website. Here are some reasons why you will like this database:
  • It contains 1186 baseline face and speech experiments
  • It is non-chimeric, meaning that the face and speech scores are actually derived from the same real person (and not chimeric)
  • The data contains controlled (clean), adverse (under challenging conditions) and degraded conditions (with different devices)
  • The baseline classifiers are very diverse, containing both generative and discriminative classifiers: MLP, GMM and SVM, as well as template-based (with correlation measures)
  • The data have been tabulated and is ready to load from Matlab easily -- all the hard work has been done for you ;)
  • It is free!
I always wanted to draft a report to publicize this database but never get to do it. If you do find this website useful, please, acknowledge by citing this website:
Meanwhile, I will let you know when the draft is ready!

Have fun!

Norman Poh


Download the score dataset here.


Please come back soon!


The original score data sets were obtained from Dr. Samy Bengio's FTP website (no longer available). The BANCA project (IST-1999-11159), an EU-funded project, is hereby acknowledged.

Norman Poh
CVSSP, SEPS, University of Surrey Guildford, Surrey, GU2 7XH, United Kingdom.