PEIPA is an archive relating to image processing and analysis, with emphasis on computer vision. Its principal aim is to provide information, datasets and software that allow the effectiveness of algorithms to be measured and compared. This is known variously as performance characterization, performance estimation and benchmarking. You should be able to see a set of links related to benchmarking in the navigation panel to your left.
Why is there this emphasis on performance characterization? A vast number of vision algorithms have been designed, and more continue to be produced, by researchers; but the choice of which algorithm should be applied under what circumstances is up the implementor, with little empirical or other evidence available as guidance. A growing number of researchers and practitioners have come to realize that the discipline will not improve in real terms until a more principled approach is adopted. Benchmarking is not the whole answer, but it is a significant step in the right direction.
PEIPA distributes test datasets that you may use to evaluate the performance of your own algorithms, and compare it with the performances of others' algorithms. The HATE test harness automates much of this process. You are welcome to upload the output files from HATE to this web-site, allowing them to be used for comparison by other algorithm developers too. The archive also has an overview of benchmarking and several tutorials -- so you can't claim that you don't know how to evaluate the performance of your own algorithms! :-)
The archive also provides information on relevant conferences, and lists of active research groups, software, and other information. You can also access back-issues of the Vision-List and Pixel mailing lists in the archive.
You can access the main parts of the archive via the navigation bar at the left of pages. PEIPA is deliberately a low-graphics site, so pages should load quickly and be usable with any web browser that supports tables. However, the test data are stored in a format that is easy to read and convert in your own software but not supported by most browsers except via plug-ins.
The PCCV Project
The performance characterization parts of this web-site are funded by an EU-funded project called Performance Characterization in Computer Vision (PCCV). The archive itself is supported by PCCV and the University of Essex. It has also received funding from and works closely with the British Machine Vision Association.
The information stored on PEIPA is, to the best of the maintainers' knowledge, correct. But be warned: they do not take any responsibility for anything you do with information, data or software retrieved from the archive; nor do any of the people who supplied them. Please notify the maintainers of any corrections or additions. General comments are also welcomed.