Four ways to make a camera understand, and when each one fits
Almost everything sold as video analytics falls into one of four families. Knowing which one you are looking at saves you comparing things that are not comparable.
Ask for three video analytics quotes and you usually get three things that are not comparable. One replaces the cameras, one adds a monthly fee, the third wants a server. Comparing prices there means nothing.
Here are the four families, with what each does well and what it costs you. We are in the last one, so read that part with the scepticism it deserves.
1. The analytics already in your recorder
Almost any modern recorder includes motion detection, line crossing or zone intrusion.
Wins when your problem is simple and perimeter-shaped: nobody should come through here at night. It is the cheapest option because you have already paid for it.
Fails when the problem is interpretation rather than detection. Somebody stealing and somebody restocking a shelf are the same movement.
2. Cameras with analytics built in
The manufacturer puts the intelligence inside the camera itself.
Wins if you are replacing the installation anyway, or fitting out a new site. Quality is high and there is no extra box.
Fails on the cost of replacing cameras that work perfectly well, and because it ties you to one manufacturer's catalogue: whatever that model does not detect, you will not be able to add.
3. Cloud platforms
The images leave your premises and are analysed on the provider's servers.
Wins on breadth of features and speed of improvement: they ship updates without touching anything of yours.
Fails in three places at once. The recurring cost grows with every camera. The connection becomes critical. And the video leaves your building, which changes your notice obligations and your risk assessment.
4. Local compute added on
A unit inside your building that analyses what your existing cameras see. This is where we are.
Wins when you have decent IP cameras already installed and you do not want video leaving the premises.
Fails if your cameras are old analogue ones or are positioned without a useful angle, because then the limit is the image rather than the analysis. Also if you have nowhere to put a box.
How to choose without comparing apples to oranges
- 01
Look at what cameras you have before asking for prices
Model, resolution, whether they are IP, and where they point. Half the options rule themselves out on that alone.
- 02
Decide whether video may leave the building
This is not a technical preference: it changes your data protection paperwork and your monthly bill.
- 03
Cost it over three years, not over the install
A small per-camera fee becomes the main line item once you multiply by cameras and by months.
- 04
Ask everyone for the same trial
One zone, one concrete question, a few weeks. Comparing rehearsed demos tells you nothing.
What this does not solve
And be wary of anyone who tells you their family always wins, us included. If you are replacing the cameras anyway, option 2 is probably a better fit than we are. We would rather say so now.
