Hi all!
Let me introduce myself. I work at the Image Processing lab of the university of Liege in Belgium. My job develop fast video segmentation and analysis algorithm that could be embedded in various devices.
Using the source code of CHDK, we were able to embed one of our algorithms and to output its result directly to the screen. Our algorithm is similar to the motion detection algorithm already present in CHDK with the notable difference that it provides a distinct detection for every pixel of the screen of the digicam (in comparison, CHDK's motion detection use a grid of something like 10x10 zones).
We made a YouTube video showing the result for those interested:
ViBe video segmentation embedded on a Canon digicam with CHDKI'm not sure that such a resolution is needed to trigger the capture of a photography but it could be used to help the auto-focus to concentrate on moving parts of the image (eg. for sport or children photography).
In any case, this shows that thanks to CHDK, computer vision researchers like us now have a portable platform on which they can embed their creations!
Thanks for the good work!
Olivier