Before you buy or deploy a video analytic system in production, it is critical that you test how many false positives it generates.

A lot of emphasis is being placed on basic tests to show an analytic can catch someone (like Bosch at ISC which I think is blatantly misleading).

I explore this in detail: http://ipvideomarket.info/review/show/15


Anyone have experiences of testing video analytics? I have done a number, some I have been happy with, others were shockingly poor.

i have a good amount of experience with analytics.
Just keep in mind that its never plug and play like they sell it to you.

Ya what would be a good test is someone standing real close to the camera,a bright light shining into the camera, while a large amount of dust or smoke partially filled the camera view also.

An analytics based system needs to be designed from the camera in, instead of designing from the recorder to the camera. Its all about your fov.
Analytics inside compared to analytics to an outside cameras is another major jump in how many false alarms you can expect.
got to run. Have a good one.