Use case
Catching illegal dumping: what the camera has to work out
Dumping is not a motion problem. It's a before-and-after problem — a vehicle stops somewhere vehicles don't stop, and the scene is different when it leaves.
VuePoint
Illegal dumping costs whoever owns the ground. Somebody drives up to a bin store, an access road, a vacant lot, or the back of a retail unit, unloads something that isn't theirs to leave, and drives off. The cost of clearing it lands on the property, and without evidence there is nobody to send it to.
The reason generic motion detection is bad at this is that the interesting part of the event is not the motion. It is the difference between the scene before the vehicle arrived and the scene after it left.
The pattern to look for
A dumping event has a recognisable shape, and it is the shape rather than any single frame that identifies it:
How the event is put together
- 1
A vehicle stops where vehicles don't normally stop
Not in a bay, not at the gate, not in the loading area. Alongside a bin store, on an access road, at the edge of a lot. Dwell time is short — typically a minute or two, long enough to unload and no longer.
- 2
Someone gets out on a route that isn't a walking route
People who belong walk between the car park and a door. A dumping event has someone moving between a vehicle and a spot on the ground, usually at the back of the vehicle, usually carrying.
- 3
The scene changes
This is the part that confirms it. Compare the ground where the vehicle stopped before it arrived and after it left. Something is there that wasn't. Nothing else in the sequence proves an event happened — a vehicle stopping briefly is not, by itself, anything.
- 4
The vehicle leaves without the load
Departure shortly after the drop, often by the same route it came in. The combination of short dwell, a person on an unusual path, and a persistent new object is what separates dumping from a delivery, a smoke break, or somebody turning around.
What the evidence has to contain to be worth anything
Knowing it happened is not the useful outcome. Being able to bill someone, or hand a municipality a complete report, is. That requires specific things to have been captured, and it drives where the cameras go.
| What you need | What it takes to capture it |
|---|---|
| The vehicle, identifiable | A camera positioned on the approach or exit at a shallow angle to the plate — not the overhead view of the drop itself. This is the single most common gap. |
| The act, unambiguous | A view of the ground where material gets left, wide enough to show the person, the vehicle, and the spot in one frame. |
| Before and after | Stills of the same area from before the arrival and after the departure. This is what makes the claim self-evident to someone who wasn't there. |
| Time and date, verifiable | Synchronised clocks across cameras and an unedited retained clip. A timestamp that can't be corroborated is easy to dispute. |
What the pattern data tells you
Individual events are useful for recovery. The aggregate is useful for making it stop. Once events are logged consistently, the same few facts tend to emerge on most sites: it clusters at particular hours, at one or two specific spots, and often around the days the legitimate bins are full.
- If it clusters right after the bins fill, the collection schedule is the cheapest fix available.
- If it clusters at one spot, that spot has an access or sightline characteristic that the others don't — usually a place you can pull in and not be seen from a window.
- If the same vehicle appears repeatedly, that changes what you're dealing with, and it changes who you report it to.
None of that is available from a folder of clips nobody reviews. It comes from events being logged in a consistent structure over months — which is the actual argument for analytics here rather than more cameras.
What this doesn't do
It detects and it documents. It does not prevent anything — someone determined to dump will dump in front of a camera, and plenty do. What changes is that afterwards you know who, when, and what, in a form somebody else has to take seriously. Whether that recovers the cost depends on your jurisdiction, your evidence, and how far you're prepared to pursue it.
Common questions
- Can it read licence plates?
- Plate capture depends on the camera, the angle, the distance, the lighting, and the speed of the vehicle — it is a placement question, not a software one. A camera positioned for plate capture on the approach will generally get it; a wide overhead view of the drop area generally will not, regardless of what is processing the image.
- What about people dumping on foot?
- The same persistence logic applies — the confirming signal is still an object that appears and stays. What you lose is the vehicle, which is usually the identifiable part, so the evidence is weaker for recovery even when the detection is just as reliable.
- How is this different from just having cameras?
- Cameras record. The difference is whether anyone looks. Most dumping on recorded sites is discovered when someone finds the pile, by which point nobody knows which of the last four days of footage to search. Detection puts the event in front of a person while the sequence is still findable.
Got a spot that keeps collecting other people's rubbish?
Tell us where it is and what the approach looks like. We'll tell you what we'd need to see to make the evidence stand up.