IPVM Team - In my world I would never use Video Analytics for a 45-90ft scene. I could just use a cheap PIR or Microwave for that. I use Video Analytics for long range applications where traditional IDS infrastructure would be too costly to install, i.e. 1,000 - 1,400ft distances. Because of this I find these tests of minimal value.
Nathan, thanks for your first comment! We have to do these tests (all of them, Axis, Dahua, Hikvision, Hanwha, etc.) to establish a baseline. We can't simply say 'trust us these suck', (1) because we need to have actual measured results, (2) how bad they are differs and (3) because lots of people do use, or at least consider, using these analytics / VMD.
I also don't disagree with your approach but many integrators will not want to take the time and money to add in an external detector, so it's important for them.
If there are other things you want us to test, feel free to share.
It’s so easy and not expensive to add an external detector and solve the intrusion issue: We make an inexpensive hardwired LX series of product that can be powered directly from some cameras and communicate the event directly to (through) the camera as well. (And a variety of other options.) You can even create “double knock” events if you have good lighting. I can’t begin to tell you how many intrusion analytics jobs we come in behind due to rain drops, spider webs, birds nests, etc with a very annoyed end user- With correct but simple planning, it’s so much more cost effective, you will have happier customers, and you provide accurate detection that can be verified through video to do it with sensors upfront instead of chasing man hours and service calls after the fact (imho).
There are a handful of cameras with PIR detectors built-in typically more residential focus. For example, I recall seeing a Costco kit offering it, like this:
Why is it not more common? I assume cost but if a Costco kit can do it, it can't be that expensive. The other issue is likely how often they would be used in professional deployments, which may be lower if the PIR range is short. Just guessing. Good question.
Yes, PIR's are/were usually on the cheaper cameras, but why not a PIR in enterprise level cameras? Would this not be more reliable than most analytics and certainly more reliable than video motion detection?
I liked the old Axis M10 series cameras that had a PIR and LED light onboard. Maximum PIR range was 20 feet, but great for dim interior coverage.
Nathan, In our campus environment we mostly record only on motion. Camera Video Analytics usually false less than camera Motion Detection. So for us and many others it's about only recording events, it's not usually used as a substitute for an outdoor intrusion detection system.
I appreciate these tests because they give a performance baseline from various manufactures and ultimately save me hard drive space.
Nathan, my experience in working with outdoor perimeter is that video analytics allow a much better stability then PIR or Microwave in terms of false alarms. The PIR/Microwave often need adjustments/service calls because of vegetation, a plastic stuck somewhere, signs moving, car lighting close-by or something else but that you can't see. With the video analytics you can actually see what is causing the false alarm and can simply mask the moving area and you're done. Analytics like Avigilon learn to eliminate many of these false alarms over time. If I look at the thousands of triggers (analytics, PIR, Microwave, radar, etc.) for a video alarm at our central, I can say that PIR/Microwave are 1st on top for the volume of false alarms. Securing a smaller perimeter (45-90ft) with PIR is indeed a much cheaper solution compared to analytics but at the end you get what you pay for :) At the end my opinion is solely applicable if the site requires remote monitoring otherwise false alarms don't really matter!
Can you expand on this question? Mainly, what camera(s) are you using? I know that some of the X series cameras are still using the ONVIF driver but most have been changed to use the Hanwha driver. I don't know if that is what you are referring to or something else. If you have the specific camera model(s), I can look into it a little more.
I am working with some Hanwha cameras and I am not completely sure about the difference between the People Counting feature under the Statistics menu and the Virtual Line from Analytics>IVA. Would any of you know?
We're considering a people counting integration into our VMS but are unsure about the difference since both of them are tripwire solutions and we could potentially get the entry/exit data from both.
To summarize, people counting does just that: counts people. IVA allows you to create a virtual line(s) to trigger events that can be pushed to record, sound an alarm, etc,.
Both do allow you to create a line, though if you are specifically looking to count people using the ability of the camera separate from a POS system/integration, then people counting would be preferred in your case.
Additionally, within "People Counting", you can enable/create reports right there in the setup page, and export files via .xlsx and tell it where to export to from there.
The Virtual lines that can be drawn through IVA allows for different directions to be used (A <- B | A -> B | A <-> B), and multiple lines cab be drawn in the same scene (Instead of just "2" for people count).
IVA can also do "Virtual Areas", which can be used for loitering, object left/placed detection. Minimum and maximum object size can be setup in IVA, whereas people counting calibration is limited to overhead "body size" so the camera has a baseline for people size.
Hope this helps, and let me know if you have any other questions!