VideoIQ will ruffle feathers with its upcoming products
VideoIQ – This analytics vendor has some amazing product that is bound
to make other analytics vendors shudder. Using an entirely different
technical approach (and different patents) than many of today’s biggest
names in intelligent video, VideoIQ conquers accuracy and speed with
the ability to pick specific people, objects, cars, etc out of a crowd. It
is self learning and when you see the demos at ASIS you’ll probably
pee in your pants. (I almost did – although it may have been all the
Peet’s Coffee Scott Schnell was feeding me). If Schnell can sell the product, it’ll give Object Video and ioimage a definite run for their money. If I were ioi or OV, I’d be approaching the videoIQ guys soon and striking a deal to play nice (no pun intended). 
I have question about technical criteria to evalaute the performance of video analytics product. who is good or bad is decided by POD and false alarm which is based efficient algorism. but another point is the detection distance. for short distance and small coverage, most system work well. for long distance (>100 meters), it will matter to the budget very much.
the distance means how many pixels you need in order to identify. meanwhie, how many pixels you can register per frame in order to extract the correct profile?