CAM-904
Covered where it matters
The junction is perfectly visible: the blue car pulling off the crossing, the red one waiting, the café with the blue awning, the corner with the yellow one, people walking along the pavement. Sharp, well exposed, nothing odd about it. And yet a third of the frame is missing: the whole left-hand strip, top to bottom, with a diagonal edge, is covered by an out-of-focus ochre mass pressed against the lens. Underneath it were the left-hand pavement and the start of the crossing. The scene has not moved, it is exactly where it always is: something is in front of it. IRIS sees that part of the field of view has been blind for a while, and that it is precisely the part it was asked to watch.
Turn on IRIS Vision
AI-generated image
What IRIS understands
What it recognises in this scene
- watched area blocked
The rule
You decide once. IRIS applies it always.
If
the area of the frame marked as important stops being visible and stays blocked beyond the margin set
Then
- alert the control room with the camera, the time, the video and which area has been blocked
- treat that area as unwatched until it clears, and say which other camera covers it
- log the time it was blocked and the time it cleared
The instruction that set it up«Tell me if the part of the frame I asked you to watch gets blocked, even when the rest of the picture looks fine.»
What changes
What changes for whoever answers for the cameras working
This is the worst of the lot, and not because it covers more but because it covers less. A black picture is spotted by anyone at a glance. This one sails through a visual check: there is street, there is colour, there is movement, so on the control-room wall it looks like a healthy camera. The day the pavement footage is needed, the pavement is not there. With IRIS the alert does not depend on the picture looking wrong: the part of the frame that actually matters — the one drawn when the camera was set up — is what gets checked, and the alert fires when that part stops being visible, however immaculate the rest is. Whoever is on call gets the camera, the time, the video and which area exactly has gone blind.
What people usually ask
How does IRIS know which part of the picture matters?
Because it is drawn. When the camera is set up, the area to be watched is marked on the picture: the door, the till, the crossing, the loading bay. That area is the reference and that is what gets checked. Outside it the picture can be blocked without anything firing, because that is not what was asked for. It is the same gesture used for counting: some things cannot be described in a sentence, they have to be traced on top of the image.
And a van that parks in front for ten minutes? Is that an alert?
It is exactly the same fact: for ten minutes that area is not being watched. IRIS says so and a person decides what to do about it. You can set how long it has to stay blocked before the alert goes out and leave it at a few minutes, so an ordinary delivery raises nothing. What should not happen is that nobody ever finds out the door has not been visible for half an hour. Raising an alert is not an accusation: it is saying that right there, right now, there is no watch.
Shouldn’t the camera itself flag this? It is a third of the picture.
The camera does not know which part of its own picture is the important one. To it, a blocked third is simply different pixels, no different from a shadow at midday or a white van parked nearby. The one who knows the crossing was underneath is the person who set the site up, and that knowledge lives in the rule, not in the lens. That is why the alert comes from the rule: the camera provides the picture, the rule provides what to look at in it.
Demo · IRIS at work
No extra alarms here. Just what needs looking at.
Not a recorded video, not an explanatory drawing: a summary of the real screen, with a different case every time it starts. There is a lot more behind it.
- The rule gets built
- It goes to every camera
- Published, it watches alone
- And then it fires
Next step
Who sees the sales that never reach the till?
10 cases: 4 raise an alert when something happens and 6 only measure. You see each image as it is, and then with what IRIS understands on top.
Who is watching when the load swings over somebody?
11 cases: 9 raise an alert when something happens and 2 only measure. You see each image as it is, and then with what IRIS understands on top.
And the best part
Behind one camera you can put a person.Behind four thousand you cannot put anyone.
Adding a camera costs little; adding another pair of eyes does not. So the cameras keep growing and the attention does not. IRIS watches all four thousand at once and flags what matters today.
Tell us your problem and we will say whether IRIS understands it, or not yet.
We reply the same working day.