Click here for Top of Page
Right Arrow: Next
Right Arrow: Previous

MUSCLE Coin Images Seibersdorf  (CIS)
Benchmark Competition 2006                                                                             

For further details please visit

muscle.prip.tuwien.ac.at

 

or contact

Michael Nölle

Video and Safety Technology

michael.noelle@arcs.ac.at

 

Michael Rubik

High Performance Image Processing Unit

ARC Seibersdorf research GmbH

A-2444 Seibersdorf

 

Allan Hanbury

PRIP, Vienna University of Technology

hanbury@prip.tuwien.ac.at

The changeover from 12 European currencies to the Euro created a unique situation. Great volumes of money had to be physically returned to the national banks of the member states. Charity organisations took the opportunity to appeal for funds. In Austria alone, the charitable donations amounted to several hundred tons of cash. Unfortunately, the coins could only be collected as a potpourri of currencies and are practically worthless unless they can be returned sorted to the national banks. The sheer volume of material rules out any attempt to separate the money manually and calls for an automatic processing device.

 

The coins originate from more than 100 countries. In general,  the material is not in mint condition. Circulation over decades has caused abrasion and dirtiness. It might be surprising that frequently the mint process itself is less accurate than one might expect. Out of centre imprints occur as well as noticeable changes of the coin die over the years. For classification purposes, more than 2000 different coin faces of over 600 different coin types were chosen, and the necessary data was collected. This defines the training data within the CIS Benchmark. The training data together with more than 60,000 coin images to be used as the test data set constitute the MUSCLE CIS-Benchmark. The data are available through the MUSCLE Benchmarking website muscle.prip.tuwien.ac.at.

 

Originally, the training data were collected for the automatic coin sorting device called Dagobert which was built at ARC Seibersdorf research GmbH and which successfully sorted the donations and thereby restored the face value of the coins.

(www.smart-systems.at/products/products_image_processing_en.html)

 

The acquired training data comprises a value in itself as it can be used to evaluate object recognition algorithms on a very large set of objects to be recognised. The test data set of the CIS Benchmark contains additional information of the true coin type and coin face for every image. This eases the performance evaluation of classification algorithms.

 

To foster the development of robust recognition and image search algorithms, MUSCLE is organising the CIS-Benchmark Competition 2006, which will make use of the CIS-Benchmark data. The call is open to researchers/groups who want to demonstrate the recognition performance of their algorithms. In order to participate, visit the CIS – Benchmark Competition 2006 website. Training protocols, formats for the results, training data and other important information is available there.

 

The results of the CIS-Benchmark competition 2006 will be presented at a MUSCLE workshop which will be held in September, 2006, together with the 28th Annual Symposium of the German Association for Pattern Recognition (DAGM06) in Berlin and gives the opportunity to present your solutions. The winning approach will be awarded a prize of €1500 sponsored by MUSCLE (although we reserve the right to split the prize in the event of a tie at first place).

MUSCLE is a European Network of Excellence that aims at fostering close collaboration between research groups in multimedia data-mining on the one hand and machine learning on the other (www.muscle-noe.org/).