Path: utzoo!utgpu!jarvis.csri.toronto.edu!mailrus!tut.cis.ohio-state.edu!ucbvax!ADS.COM!Vision-List-Request From: Vision-List-Request@ADS.COM (Vision-List moderator Phil Kahn) Newsgroups: comp.ai.vision Subject: Vision-List delayed redistribution Message-ID: <8902150225.AA13600@deimos.ads.com> Date: 11 Feb 89 00:54:57 GMT Sender: daemon@ucbvax.BERKELEY.EDU Reply-To: Vision-List@ADS.COM Distribution: inet Organization: The Internet Lines: 205 Approved: vision-list@ads.com Vision-List Digest Fri Feb 10 16:54:57 PDT 89 - Send submissions to Vision-List@ADS.COM - Send requests for list membership to Vision-List-Request@ADS.COM Today's Topics: - Machine Vision and Applications, Volume 2, Issue 1 - Research Fellow in Computer Vision - RE: NSF Request for Industrial Research Topics ---------------------------------------------------------------------- Date: Thu, 09 Feb 89 11:32:58 PST From: Gerhard Rossbach Subject: Machine Vision and Applications, Volume 2, Issue 1 There is now a journal in the field of machine vision, integrating theory and applications. "Machine Vision and Applications", the international journal from Springer-Verlag is now in its second year. It is published four times a year and has a personal subscription rate of $45.00 (including postage and handling). The institutional rate is $105.00 (including postage and handling). Volume 2, Issue 1 will be published the beginning of March, 1989. The contents for this new issue are: "Performance Assessment of Near Perfect Machines," Robert M. Haralick. "Combined Decision Theoretic and Syntactic Approach to Image Segmentation," by W. E. Blanz and B. J. Straub. "Real-Time Model-Based Geometric Reasoning for Vision Guided Navigation," by Uma Kant Sharma and Darwin Kuan. "Report on Range Image Understanding Workshop, East Lansing, MI, March 21-23, 1988," by Ramesh Jain and Anil K. Jain. For further information on submitting a paper or subscribing to the journal, please send email to: rossbach@hub.ucsb.edu or write to Springer-Verlag, 815 De La Vina Street, Santa Barbara, CA 93101 ------------------------------ Date: 10 Feb 89 09:47:00 WET From: JOHN ILLINGWORTH Subject: Research Fellow in Computer Vision University of Surrey, Department of Electronic and Electrical Engineering, United Kindom. ********************************** RESEARCH FELLOWS : COMPUTER VISION ********************************** Two Research Fellows are required in the Department of Electronic and Electrical Engineering for projects in Computer Vision. The project is called Vision As Process, VAP, and is funded under the ESPRIT BASIC RESEARCH ACTION, BRA, program. It is an international collaboration involving institutes at Linkoeping and Stockholm (Sweden), Aalborg (Denmark) and Grenoble (France). Surrey Universities major contribution will be the development of a high-level scene interpretation module. The project will be carried out within an active research group in Vision, Speech and Signal Processing. The group comprises about 20 members and has extensive computing resources including SUN, VAX and Masscomp computers as well as specialised image processing facilities. The sucessful candidates will be required to develop, implement in software and experimentally evaluate computer vision algorithms. Applicants for these post should have a degree in mathematics, electronics, computer science or physics. Previous experience in computer vision, image analysis, knowledge based methods or pattern recognition will be an advantage. The appointments will be initially for two years with a salary in the range 9,865 to 14,500 UK pounds per annum depending upon age, qualifications and experience (superannuation under USS conditions). Applications in the form of a curriculum vitae (3 copies) including the names and addresses of two referees should be sent to the Personnel Office (JLG), University of Surrey, Guildford, Surrey GU2 5XH. Further information may be obtained from: Dr J. Kittler (0483 509294) or Dr J. Illingworth (0483 571281 ext 2299) at Department of Electronic and Electrical Engineering, Surrey University, UK. ------------------------------ Date: Tue, 07 Feb 89 11:29:23 -0500 Subject: RE: Request for Industrial Research Topics From: "Kenneth I. Laws" > From: skrzypek@CS.UCLA.EDU (Dr Josef Skrzypek) >>> MOST ACADEMICS START WITH TOOLS AND DO RESEARCH TO FIND APPROPRIATE >>> PROBLEMS AND FUNDING AGENCIES.... > This is a rather naive view of academics approach to problem solving. > Ken, you have mixed it up. Normally it's the other way around. One > thing is certain, having tools and poking around, with the hope of > finding a problem is a prescription for very poor science and probably > bad engineering. Is there a fundamental change in NSF philosophy? No, NSF hasn't changed -- except for the constant turnover in program directors and their viewpoints. And I'll admit to overstating the case. I don't know the proportion of academics who start with tools, and I have certainly seen engineers (or at least members of technical staff) who have wanted to start at the tool end. Witness the recent burst of activity in neural networks, or previous interest in orthogonal transforms, Hough transforms, iterative relaxation, pyramid relaxation, etc. I share your concern that this leads to bad science (and especially to bad engineering). In a few cases, the emphasis on tools is wholly justified. Someone has to study them, and to provide expertise to those who need the tools to solve real problems. Mathematics and other core sciences have always nurtured such people. NSF is now charged with supporting engineering research and tech transfer, as well as with traditional support of science and of engineering education. This broad mandate motivates us to ask about relevance in funded research. We would like to see progress toward long-range goals of national importance. Unfortunately, few such goals have been identified. The value of research in computer vision is fairly obvious. After all, how many industries could expect to remain competitive in world markets if they hired only unsighted workers? The value of research in mobile robotics or dexterous hands is less clear, and I therefore expect stronger justification in such proposals. For academics working on real-world problems, this should not be difficult. Very often, however, a professor whose expertise is at the tool end depends on his graduate students to prove theorems. The students are thus trained mainly for academic careers, and may even fail there if they cannot locate customers willing to support such research. Thus the desperate search for applications, as well as for grants. Unfortunately, many of these people drop out of the science/engineering pipeline before finding the support they need. Engineers rarely fall into such a trap. There are cases where someone's expertise becomes outdated and the problems he knows how to solve are no longer problems. This can happen in pattern recognition, for instance, when a system functions well enough that there is no point in further improvements. Still, an engineer who has solved one problem can usually find work solving another. There is less of a tendency to stick with just the tools that one has used in the past, more of a tendency to search for tools appropriate to the application. At NSF, we commonly deal with proposals about applying a particular tool to a particular problem. The need for the research is often justified in terms of the problem, but the scientific merit is usually judged by what it will teach us about the tool. We try to balance the two to best serve the nation, but our review process and funding policies typically favor the tool builders. The particular problem that I was bringing up, and for which there have been no responses, is the need for a list of research goals for our Small Business program. Or for any program, for that matter. The engineering directorate is generally able to point to specific problems of national importance that they are trying to solve. The computer science directorate has more difficulty with this. We talk about bigger, smaller, faster, cheaper, more robust -- but what are these computers and algorithms really need for? The COSERS report was one attempt to answer this. Our small-business people need more-specific projects, however. As do the nation's graduate students. One way to get a handle on the problems is to ask about in-house research efforts that have failed. Perhaps someone else could solve these problems, avoiding whatever technical or personnel difficulties arose at your site. Or perhaps someone in management has been poking around saying "We need X. Can we do it in-house? Is there a supplier? Who do we call?" Have you ever had a blue-sky proposal that never got off the ground because the vision technology just wasn't there? Have customers called, asking for something that just wasn't your product line? Or, if you'd rather brainstorm, I'm open for blue-sky suggestions. Over the next decade, my program may be pumping something like $50M into machine perception. How would you like to see that money spent? What should our national goals be? Which research community should be supported? Where should we be one or two decades from now? What would we do with ultimate vision systems if we had them? What's keeping us from doing it now? Any inputs? -- Ken Laws ------------------------------ End of VISION-LIST ********************