False Positives outweigh your catch rate?

[jessecf]

I've been testing analytic systems for over a year now on some projects that I'm working on for some clients in NY. Has anyone had extremely good success with a particular analytic company?

What I'm finding is that almost always I can get the detection I am looking for, but the false positives are extremely high even after tuning the product algorithms and camera views. I've been working on perimeter intrusion, loitering, UAO, stopped vehicle.

On 7th ave in NY we were doing a test for UAO outside of our building. We did catch the bag that was left. We also received about 400 false alarms in less than an hour. Just an example of a busier area we are currently testing.

[Wes]

Welcome to the site!

I have tested a few video analytics software packages and found that there are still false alarms as well. In my opinion, the video analytics industry is still very new as well as the software. I have used the software for various installations that require a higher level of motion detection and it does work well. I wouldn't say it works at 100%. They still get false alarms but it is still a lot less than standard motion detection.

Within the next few weeks I'm going to begin doing some heavy reviews on all the video analytics software available. I will be posting the reviews on the website. It will also give me a better understanding of each software company allowing me to give better product recommendations.
_________________
Wes
VideoAnalytics.net
www.videoanalytics.net

[funkypixels]

You're going to need something with background learning and change detection. A pure VMD system will always struggle with a busy scene like 7th Av!

Trouble is most of these analytics systems grew out of the military and therefore are best suited to guarding sterile areas, not busy areas.