Obvious Vs Complex

I am in New York at PerCom, where I will be giving a Keynote on eventweb.  This conference is being featured at Seraja.  An interesting experiment.  All aspects of the conference are covered on Seraja.

Yesterday I had dinner with Arun and Kusuma Hampapur.  Arun developed a surveillance system at IBM which is now being installed by IBM at several places.  This is becoming a strong offering by IBM.  Very exciting to see that a computer vision based system will be really deployed at mass scale. Good for Arun and Chiao-Fe Shu who developed the system.  But it is better for computer vision as a field.

In discussion talk went to papers that I wrote in 1976 time frame of use of difference pictures on video analysis. That work was done in Prof. Nagel’s lab under his guidance at University of Hamburg.  Those difference pictures became the basis of so many video based computer vision systems that even I am surprised.  Differencing has become the first step in many practical computer vision systems used to detect changes and do motion analysis.  I still remember being very excited by those results.  Nice to see that such a simple idea has resulted in so much use. 

In science and technology, many times researchers are impressed by compex processes rather than simple approaches.  I find it intriguing how many times researchers ignore simple straight approach in favor of mathematically impressive and complex approaches and then get frustrated that those sophisticated approaches don’t give desired results.  Only later they find — and find it difficult to accept — that the siple approach is in fact the effective one. 

I guess, it is easy to ignore the obvious.