In 1982, if you wanted to know about a technique or idea from computer vision, you just reached for your Ballard and Brown. Probably you still do, but just wish that it were more up-to-date. If you are a teacher, you will be constantly searching for a reasonably priced book to use for your courses, only to find that most authoritative books are too expensive for anything other than libraries, and most reasonably priced books are rather thin and one-sided. And if you find one, it is inevitably out of print and/or out of date in only a few years. Academic staff often find that the only solution is to write their own notes - often a long, frustrating and sometimes unrewarding activity.
With the WWW, there is a remedy for these problems. Edinburgh University presents: CVonline: The Evolving, Distributed, Non-Proprietary, On-Line Compendium of Computer Vision
The Compendium is a collection of hypertext summaries on the central topics in computer vision. We have organized an index of about 700 topics. (The top-level categories are listed at the bottom of this message.) Everyone will have written some notes on a few topics - we propose to set up the WWW links from the central index to your text. Thus, everyone can benefit from your efforts, and the computer vision community can advance by having a common set of widely read text materials available at virtually no expense to your library nor students.
The Compendium is intended to be a clear summary of methods and applications of computer vision, organized into sections covering the main topics of practice and research. Each section will contain a number of topics, and each topic will be a hyperlink to a set of materials associated with that text. The entry point into the Compendium is: http://homepages.inf.ed.ac.uk/rbf/CVonline
How can the Compendium be used?
We are not being prescriptive about what material should be taught. Instead, the Compendium can help provide the materials to help you teach it. You need only link to particular topics in the Compendium and then you will not need to write the notes yourself. You can choose the topics and their order of presentation. All of the topics should be in one place and it will be identifiable as to which have text ready for use.
To improve access, there will be shadow sites in the US, Asia, continental Europe and elsewhere according to demand.
What is the Compendium status?
By mid 2007, there were about 3000 contributions covering about 1400 of the the 1600 topics in Cvonline.
Why is CVonline realistic and necessary?
We estimate that the amount of material generated in the image processing and computer vision community since 1980 would fill between 10 and 100 "Ballard and Brown"s. No one could possibly be able to or even be interested in writing this amount of text. This is where the distributed aspect comes in.
Who can make a contribution?
There are probably 1000 instructors of computer vision topics in the world, and probably 5000 graduate students eager to make their mark. Perhaps you have a relevant section from some course notes, or have had to write a survey and summary, that you'd like to contribute? On the average, everyone only needs to contribute a submission on only one topic to make the Compendium work.
Of course, initially the Compendium will be fragmentary, suffer from inconsistent style and notation, have some poor quality articles, be incomplete and evolving. But, if an article is weak, someone will eventually get fed up and submit a better article. This also means that the Compendium is also always nearly up-to-date: new techniques, evaluations and comparisons will keep being added over the years.
What do the contributors get?
Why should you contribute your effort to this project instead of writing your own book?
Our experience is that, for most authors of computer vision books, the process of writing is intellectually rewarding, but not particularly financially rewarding, resulting generally in a few thousand dollars of royalties for a thousand hours of work. If you want to make money, there are more effective methods. This is not to say that there is no place for coherent, in depth, studies nor monographs on special topics. A general reference does not replace a specialized book.
Is there a financial benefit to Edinburgh University (or the people there)?
Not that we know of. If successful, we might get some notoriety and perhaps a promotion.
© 2007 robert fisher