OAGM: PATTERN RECOGNITION ACTIVITIES SINCE 20 YEARS
A retrospect and outlook of pattern recognition activities in Austria, presented by Prof. Franz Leberl at the occasion of the 25th OAGM Meeting in Berchtesgarden.
Abstract
The Austrian Association for Pattern Recognition AAPR/OAGM has been very successful in maintaining 20 years of exciting, youthful and internationally attractive activities, reflected in 25 scientific meetings and their printed proceedings to record these meetings. The following reflects on the development of OAGM and comments on what could be an even more exciting future that can build on a very successful past.
OAGM is very much alive!
It is gratifying to be able to attend the 25th meeting of the Austrian Association for Pattern Recognition (or OeAGM) with so many attendees. Of particular importance to me is the fact that the average age of the attendees is low, many have recently graduated and are pursuing a doctorate. We can expect that the OeAGM will continue to do well in the future.
We have here a meeting of about 75 people. Let me explain why this must be considered a large number. The German counterpart is the Deutsche Arbeitsgemeinschaft für Mustererkennung (DAGM). Given Germany’s size at 10 times that of Austria, the DAGM-meetings would have to attract 750 attendees. The world-wide umbrella organization is the International Association for Pattern Recognition (IAPR); its conferences would have to attract 100 times more people than Austria alone, thus 7,500 attendees to reflect the Austrian level of activities, when in fact the last International Conference for Pattern Recognition (ICPR) had about 1200 attendees.
So I cannot but be proud about this Austrian success.
A look at the beginning
I may well be the oldest participant in this meeting. I am at least the only one in attendance today who was present when the OeAGM was formed. In 1981, I tried to interest the Austrian Computer Society (Österreichische Computergesellschaft OCG), under its President Dr. N. Rozsenich, to support a new chapter (Arbeitsgruppe) for pattern recognition. Other chapters existed at the time, for example for Computer Graphics. The proposal for the new Arbeitsgruppe was made at the annual OCG-meeting in early 1981, was approved and some limited funding was promised for the publication of proceedings, for the invitation of speakers, for mailings etc.
So later in 1981, we created the Österreichische Arbeitsgruppe für Mustererkennung (OeAGM) in a start-up meeting at the University of Salzburg, in the office of Dr. G. Bernroider. The founding “fathers” were Dr. Bernoider (Zoologie), Prof. F. Pichler (Systemtechnik, Univ. of Linz), Prof. N. Albrecht (Mathematik, Univ. of Innsbruck) and myself, then an a.o. Professor in Photogrammetry and Remote Sensing at Graz University of Technology’s School of Civil Engineering.
Although Vienna is the home of the OCG, we missed Vienna in our team. This was fixed by creating an affiliation with the Studiengesellschaft für Kybernetik, headed by Prof. R. Trappl of Vienna University, and by an invitation to Prof. Schuster, who was involved in medical imaging at Vienna University, to join our group.
We needed a chairman of the OeAGM. I was elected to a 2-year stint as Chairman (Vorsitzender). That appointment was extended for another 2 years in 1983. One of the reasons was that my credentials included the recent founding of the Institute for Digital Image Processing and Graphics of Joanneum Research, and this appeared as the entity in Austria that was most clearly focussed on image processing and pattern recognition, thus the core topics of OeAGM. When I dropped out from academia in early 1984, a new President was needed and it was very fortunate that the function was taken over by Prof. W. Kropatsch, who then was still employed at Joanneum Research in Graz.
At the founding meeting in Salzburg, we agreed to have an inaugural workshop in Ramsau later that year. The title page of the proceedings of that meeting is shown in Figure 1. We had 14 papers and 32 attendees. At the time, a number of people attended from industry. We went to the trouble of recording the discussions on tape and then to transcribe the discussion. The table of contents reflects subjects that are still of interest today (see Figure 2).
While the topics of the papers were very diverse, reflecting the diverse interests of the attendees, we all considered this first meeting as very stimulating and had therefore a number of follow-on workshops, sometimes two per year. As a result we now have an anniversary of 20 years for the organization, but the 25th OeAGM-meeting.
The initial years of the OeAGM were as an Arbeitsgruppe of the OCG, and as the national partner of the International Association for Pattern Recognition IAPR. This led to me being invited to take a position in the directorate of that association and to run one of the 12 or so Technical Committees of IAPR. This put Austria on the map in the international pattern recognition scene, and created many of the scientific relationships from which many among us still benefit to this day. We also developed a very friendly relationship with the DAGM. This very soon led to the DAGM-annual meeting being held jointly with the OeAGM-meeting in 1984 (Graz) and 1994 (Vienna).
As will probably be explained elsewhere, the OeAGM changed from a mere Arbeitsgruppe of the OCG to a separate society (“Verein”), still affiliated with the OCG. But this was already under its new leadership and after my emigration to the USA.
Pattern Recognition?
One might ask why the focus and choice of name was “Pattern Recognition”? After all, we have today many different concepts that all somehow define very similar fields:
1. Image Processing
2. Image Understanding
3. Image Analysis
4. Image Pattern Recognition
5. Computer Vision
6. Machine Vision
7. Robot Vision
8. Industrial Vision
The answer is simple. At the time we first had the desire to address not only images but also other sensor data such as sound. Second we were inspired by the International Association for Pattern Recognition (IAPR) with its very well-known and respected founding fathers Azriel Rosenfeld, Herb Freeman, Theo Pavlides and others, and we wanted to be an IAPR national affiliate. Third, we had the role model of the DAGM. We were unaware of any other international societies we could have affiliated ourselves with. Today of course there would be others, for example within the IEEE-organization (Medical Imaging, Computer Vision….).
On reflection, I think that we did the right thing. “Pattern Recognition” is a good umbrella for the scientific work of OeAGM-members, and the de-facto focus on “images”, and thus on image pattern recognition, is not really a contradiction.
Characterizing the current constituency (Status 2001)
A few things have changed over the years. Initially, the OeAGM, very much like the OCG, was meant to be both academic and industrial and to draw people together from both arenas and backgrounds. The founding member’s affiliations may have been at the Universities of Linz, Innsbruck, Salzburg and the Technical University Graz, but it was the then new Institute for Digital Image Processing and Graphics (founded the previous year in 1980) at Joanneum Research which provided the initial organizational backbone to OeAGM. This may have caused an orientation towards the applications.
With the creation of the only Austrian Professorship for Pattern Recognition at Vienna Technical University, that Institute became the driving force of OeAGM. As a result, the OeAGM became more academic, and we see the effect represented in the audience of today’s 25th meeting.
Today, there are probably two or three academic institutions that “carry” the OeAGM. Most prominently of course is still the Pattern Recognition group at Vienna University of Technology (Prof. Kropatsch) and its sister at the Institute for Computer Graphics and Vision at Graz University of Technology. A few additional University-groups include the current OeAGM president’s institute in Graz. The growing number of non-University research centers and research companies in nearly all Austrian provinces should produce a noticeable diversification of this support.
About future directions (Status 2011?)
Let me address the future with a few thoughts. I believe that there is the potential for OeAGM’s growth in both size and significance, and that this potential should be addressed and realized.
First, it would seem to me that the OeAGM should understand itself not only as a learned society of academics, but also as a “lobby” for more and better academic research. Initially, computer vision and image analysis were clearly applications of computers and not considered research fields in computer science. That has slowly changed and today, making a computer “see” has become one of the central themes of research in computer science and information technology, with significant effects on industry and international trade. The OeAGM can and should be heard in the media and in politics on these developments. The OeAGM should write position papers on policies.
Second, the OeAGM might want to reconsider its focus on “research” and consciously address the industrial applications as well. If successful, this would draw in more interest and participation from industry, with hopefully multiple applications tracks. Industrial concerns should be encouraged to sponsor OeAGM events, should provide fellowships for academic training positions, should sponsor academic chairs. Currently, two Austrian K-Plus cooperative programs between academia and industry are in operation (these are two from a total of twelve), with a focus on images in computers. One center is called “Advanced Computer Vision”, the other “Virtual Reality and Visualization”. This is significant since it documents the industry’s need for know-how in digitally processing visual information.
Third, I argue that the OeAGM-focus on processing of sensor-produced images is too narrow. I suggest that the wall between computer vision and computer graphics be broken down. OeAGM should understand itself as being an organization to process visual and other sensory information in computers. More and more examples can be listed where computer graphics is no longer a topic separate from computer vision. By inviting keynote speakers, by sponsoring fellowships, by awarding prizes for work done on specific topics, OeAGM should consciously broaden its scope to include computer graphics, speech and signal processing, digital interactive television, virtual and augmented reality etc.
Fourth, the traditional affiliation with IAPR should not be the only association. IEEE, SPIE, Eurographics and other organizations behind European and international conferences exist. They all should be on OeAGM’s radar screen, and OeAGM should position itself as the Austrian affiliate of these organizations. If that affiliation is already occupied by an existing but not very visible entity, a merger with that entity should perhaps be considered.
Conclusion
The future looks bright. OeAGM’s 20-year history tells a fairly unique story, not easily duplicated by other scientific themes in Austria. This success should be the basis for more! It should encourage young people to get involved, to make their own mark, and to create a society that is even more successful, larger and more influential than today.
This 25th OeAGM-meeting is being organized by one of the bright young stars of the Austrian computer vision scene, Dr. Stefan Scherer. It is being held in his home town of Berchtesgaden. This is a beautiful place quite worth a visit even without the excuse of a scientific meeting, so much so that I promise to be back with the family. Given his skill, I know that we will have a very successful meeting!