Computer vision

Computer Vision is a field of study that studies visual processes to understand recovery of information from one or more images of a scene. This understanding can then be used to develop useful application systems to extract meaningful information. These systems have applications in many different domains ranging from medical diagnosis to homeland security. These techniques will be the foundation behind many emerging multimedia applications including image and video databases.

One of the most prestigious academic conference in this field is CVPR (Computer Vision and Pattern Recognition). This conference was in San Diego for 2005 in the last three days – 22-24th June. I did research in this field from about 1975 to 1993 very actively. This was my major research focus. In fact I have been involved in this field from its early stages. Prof. Azriel Rosenfeld, a father figure in this field, was the person who suggest that I come to USA when I met him in Germany in Prof. H.H. Nagel’s lab where I was a post-doctoral researcher. I worked on so many interesting problems in this area. I must say that much of my professional enjoyments and accomplishment in my formative professional life came in this field.

Computer vision is a truly interdisciplinary area. People from many areas ranging from mathematics to neurosciences; electrical engineering to cognitive scientists contribute to this area. This has been the strength of this field. But this has also been the weakness. The field has never been able to develop as a technology or as a science. It could become a synergistic combination of the science and engineering, but has not been able to clearly get in any of these two and has remained kind of both. It was interesting to hear that field is growing so much so that there were more than 1000 attendees at this conference – a very good number for an academic conference. But there are not enough positions for students finishing their PhD in this area. People were saying that they are seeing good signs because many big companies are hiring in this area. I find this intriguing. This should be one of the most applied field. People are putting camera everywhere from our pockets to all highways. We deal with images in everyday medical diagnostics and video surveillance is everywhere. All these require understanding of computer vision. And one would think that this field should be in great demand.

This may in fact be a very interesting example how academia has evolved. Knowledge creation is an important task in academic research. In many areas, however, knowledge is created just for its own sake without any consideration of real applications or consideration of reality. Since reward system in academia is number of papers published in high quality journals, the goal of researchers becomes publishing papers. Knowledge creation or problem solving is not the goal; writing papers is.

Computer vision has made good progress but it is still far from where it could have been. I feel that this field should be one of the core fields in computer science because it is at the center of dealing with computational processes leading to extraction, organization, storage, application, and presentation of visual information. And more than two thirds of all information that we deal with comes in visual form. A decade ago technology was not ready, but now it is. A good leadership in this field could really make this a central theme in computer science.