Pick what someone will act on
An alert nobody has a response for becomes noise within a week, and noise trains operators to ignore the screen.
Use cases
Video analytics use cases, and what each one needs
Twenty-two things video analytics can do across safety, compliance, traffic and operations — with the camera conditions each one depends on.
Home › Video Analytics Use Cases
Use cases
22 video analytics use cases, and what each one needs
Every vendor publishes a use-case list. Very few tell you what has to be true of the camera for it to work. The third column is the one worth reading — it is where most disappointing deployments were decided.
Safety and securityCompliance and processTraffic and vehiclesBusiness and operations
Category
Safety and security
| Use case | What it does | What it needs to work |
|---|---|---|
| Perimeter intrusion | A line or zone crossed after hours raises an alert with the clip attached. | Camera must see the whole approach, not just the fence line. |
| Loitering | A person staying in a zone longer than a set time — the pattern before most incidents. | Needs a stable view; a swaying camera generates false alerts. |
| Abandoned object | A bag or package left and not collected within a set window. | Works in clean spaces; a cluttered warehouse floor confuses it. |
| Crowd density | Continuous headcount per zone with a threshold alert before it becomes a crush. | Overhead or high-angle mounting; eye-level views under-count badly. |
| Wrong-way movement | Anyone or anything moving against the expected direction in a corridor or lane. | Direction has to be genuinely one-way for this to be meaningful. |
| Fall detection | A person going down and not getting up, in areas where someone may be working alone. | Needs a clear floor view; furniture occlusion breaks it. |
Category
Compliance and process
| Use case | What it does | What it needs to work |
|---|---|---|
| PPE compliance | Helmet, vest or mask missing on someone entering a controlled zone. | Camera at the entry point, close enough to resolve the head and torso. |
| Restricted zone entry | Anyone entering a zone they are not authorised for, during defined hours. | Pairs best with access control so authorised entries do not alert. |
| Occupancy limits | People in and people out, with a live count against a stated maximum. | A dedicated overhead counting camera at each door, not a corner view. |
| Housekeeping and spills | A zone that should be clear but is not, checked on a schedule rather than continuously. | Better as a periodic check than a live alert — fewer false positives. |
| Idle equipment | Machinery stationary during a shift when it should be running. | Simple and reliable; one of the fastest to deploy. |
Category
Traffic and vehicles
| Use case | What it does | What it needs to work |
|---|---|---|
| ANPR number plate capture | Every plate logged with a crop, searchable across months and junctions. | Dedicated ANPR camera down the lane, IR at night. |
| Red light violation | Vehicle crossing the stop line against the signal, with plate and phase recorded. | Needs the signal phase as an input, not just video. |
| Helmet violation | Two-wheeler riders without helmets, captured with the plate. | Front-facing camera at rider head height, good daylight. |
| Speed estimation | Average speed between two points, or spot speed where geometry allows. | Requires calibration on site; uncalibrated speed is not defensible. |
| Stopped vehicle | A vehicle stationary on a carriageway or in a tunnel where it should not be. | Critical in tunnels — pair with the ventilation and signage system. |
| Parking occupancy | Bay-level free and occupied counts for a lot or a street stretch. | One camera can cover several bays if mounted high enough. |
Category
Business and operations
| Use case | What it does | What it needs to work |
|---|---|---|
| Footfall counting | People entering and leaving, by hour, day and door. | Overhead camera at the door; this is the single biggest accuracy factor. |
| Dwell time | How long people spend in a zone — an aisle, a display, a counter. | Needs zones drawn to match how the space is actually used. |
| Queue length and wait | Live queue depth with an alert when it crosses your service threshold. | Works where the queue forms in a predictable place. |
| Heatmaps | Where people go and where they never do, aggregated over days or weeks. | Useful as a trend, not as a precise number. |
| Shelf availability | Gaps on a shelf flagged for restocking. | Camera per bay section; the most camera-hungry use case on this list. |
Choosing
How to pick the first two
Sites that start with one or two analytics and expand succeed. Sites that switch on twelve at once end up ignoring all of them.
Check the camera first
Most failures are a mounting decision, not a model. If the camera is wrong for the job, no software fixes it.
Measure the before
Record what the current process misses for two weeks. Without that number you cannot tell whether the system helped.
Tell us the problem, not the feature
Describe what you are trying to catch or count. We will tell you which of these fits, what it needs, and whether your cameras can do it.