It has been a while since I have mentioned John Honovich and
his IP Video Market web site (www.ipvm.com).
Long time readers might recall that John’s site is a very informative location
for unbiased information on all things associated with video surveillance. He
recently posted
a report (membership required, sorry) looking at comparing the operations
of three different video analytic systems in the rain.
Most modern facility security systems use some sort of video
surveillance system. Larger installations are shifting to video analytics to
reduce the manpower necessary to adequately track outputs from a large number
of video cameras. A proper assessment of the adequacy of such a system must not
only take into account standard daylight and night-time operations but also a
wide variety of expected weather conditions.
The IPVM.com report provides some interesting test results
for detecting both personnel and vehicle line crossing (a typical video
analytics task) at ranges from 20 feet to over 200 feet from the camera in
moderate to heavy rain. The tests were conducted in a parking lot so a lot of
the issues associated with vegetation in the target area were eliminated.
While the missed target data on the three systems tested
vary substantially, I was really surprised that their test did not have a
single false positive. John’s tester did not mention this in the report, but I
wonder if the sensitivity of the system shouldn’t have been tweaked up just a
bit; an occasional false positive in these adverse conditions might have
increased the target detection rate.
What is important in John’s report is that target detection
distances are significantly impaired by rain events. Any infantryman knows that
rainy days favor the attacker and this just goes to show that that fact is not
just due to cold, wet, miserable folks being less than attentive. Even video
surveillance systems are affected.
Posts
No comments:
Post a Comment