How Cities Monitor Traffic: The 5 Methods, and What Each One Misses
Cities monitor traffic using loops, cameras, radar, manual surveys and probe data. Here's how each method works, what it costs, and what it leaves out.
Most Indian cities monitor traffic on a few hundred metres of road and guess about the rest.
That is not a criticism of the people running them. It is arithmetic. A camera covers the junction it is pointed at. A city has thousands of kilometres of road and a few hundred cameras, so roughly 97% of its roads are invisible to the monitoring network. The question every traffic authority eventually runs into is how to manage the roads nobody is watching.
This is a guide to how cities monitor traffic today: the five methods in use, what each one actually measures, what it costs, and the blind spot each one leaves behind. If you are evaluating a traffic monitoring system, the blind spots matter more than the feature lists.
The five traffic monitoring methods
1. Inductive loop detectors
Wire loops cut into the road surface. When a vehicle passes over, the metal changes the loop's inductance and the system registers a vehicle.
What it measures: vehicle count and presence, at that exact spot.
Where it works: signal actuation. Loops are how a signal knows a vehicle is waiting.
The blind spot: a loop knows a car crossed a line. It does not know where that car came from, where it is going, or how long it sat in a queue 300 metres back. And loops fail. They get dug up during road work, waterlogged in monsoon, and quietly stop reporting. A dead loop looks identical to an empty road.
2. CCTV and video analytics cameras
Cameras at junctions, increasingly with AI on top that counts vehicles, classifies them, and flags incidents.
What it measures: everything in frame, counts, classification, queue length, red-light violations, ANPR.
Where it works: enforcement and junction-level operations. This is the richest data of any method, inside the frame.
The blind spot: the frame. A camera sees 50 to 150 metres of road. Coverage costs scale linearly. Want to watch another junction? Buy another camera, run power to it, run backhaul, maintain it. Cameras also degrade: lens fogging, misalignment after a storm, power cuts, backhaul failures.
3. Radar and LiDAR sensors
Radar bounces a signal off vehicles to measure speed and presence. LiDAR builds a 3D picture with laser pulses.
What it measures: speed, count, and position, reliably in fog, rain, and darkness where cameras struggle.
Where it works: speed enforcement and highway monitoring.
The blind spot: same as cameras. It is a point sensor with a range. It tells you what happened at one location. Radar also has a known failure mode where a single vehicle gets counted in two lanes.
4. Manual traffic surveys
People with clipboards, or a consultant hired to run a study, counting vehicles at a junction for a few days.
What it measures: whatever the surveyor was asked to measure.
Where it works: one-off planning studies, before-and-after checks on a specific intervention.
The blind spot: a survey captures three days. A three-day count in February tells you nothing about a Ganesh Chaturthi Friday, or about what happened on that corridor last Tuesday when the flyover was shut. Continuous city-wide surveying is not affordable, which is the entire constraint.
5. Probe data from vehicles
Anonymised, aggregated location data from the phones and navigation apps already in millions of vehicles. Google's Roads Management Insights (RMI) feed is the version most Indian cities will encounter.
What it measures: actual speed and travel time on road segments, across the whole network, continuously.
Where it works: network-wide congestion, speed trends, corridor performance, and the before-and-after impact of an intervention.
The blind spot: it is honest about what it cannot do. Probe data will not read a number plate, will not issue a challan, and will not tell you a truck was in the left lane. It measures how traffic is moving, not who is in it. If your problem is enforcement, this is the wrong tool.
Why coverage is the question that actually decides things
Put the five methods side by side and one number separates them.
| Method | What it sees | Coverage |
|---|---|---|
| Inductive loops | Vehicle presence at a point | The loop |
| CCTV / video analytics | Rich detail, in frame | 50–150 m per camera |
| Radar / LiDAR | Speed and count at a point | Sensor range |
| Manual surveys | Whatever was counted | One site, a few days |
| Probe data | Speed and travel time | Every road with vehicles on it |
The first four are point methods. They answer what is happening here. Add more of them and you get more points, and a bigger bill, and more devices to maintain.
Probe data is a network method. It answers what is happening across the city, including the arterial road nobody put a camera on, because vehicles are already driving down it with phones inside them.
A city running only point sensors can tell you precisely what happened at 40 junctions and nothing at all about the roads between them. That is the situation most traffic authorities are in, and it is why the honest answer to "how bad is congestion in our city" is usually a shrug and a number from a survey somebody ran two years ago.
The five methods are not competing for one slot. They answer different questions:
- Enforcement: you need cameras. Nothing else reads a number plate.
- Signal actuation: loops or radar at the junction.
- Network-wide congestion, speed, and corridor performance: probe data. Point sensors physically cannot answer this.
The mistake is buying method 2 and expecting it to answer the method 5 question. A city can invest 50+ crore in an ITMS and still have no measurement of average speed across most of its road network. Cameras report on the junctions they are mounted at. They were never pointed at the other 97% of the roads.
In Pune, average vehicular speed across the city's major corridors rose from 20 km/h to 26.8 km/h over two months of use, a 34% gain. That figure is not a vendor estimate. Pune City Traffic Police reported it themselves:
"Over the last two months of using the application, we have observed that the average vehicular speed has increased to 26.8 kmph from 20 kmph."
Additional Commissioner of Police Manoj Patil, Pune City Traffic Police
The measurement covers 964 roads. A camera network could not have produced it, because the cameras were never pointed at most of those roads. The data was already there, in the vehicles. Read the full Pune case study →
Cameras and probe data answer different questions
Cameras and probe data are not rivals, and most cities should run both. Cameras handle enforcement and junction operations. Probe data handles network-wide congestion and speed, the 97% of roads no camera was ever pointed at. See what TraffiCure measures across a city, and how it compares with a traditional ITMS.
See what your city's roads actually look like. TraffiCure measures speed and congestion across every road in your city, with no cameras to install. Book a demo →
Frequently asked questions
How do cities monitor traffic without cameras?
Through probe data, anonymised and aggregated location data from vehicles already on the road. It gives speed and travel time on every road segment with traffic on it, with no hardware to install. It cannot do enforcement, but for congestion and speed measurement it covers the whole network rather than a few junctions.
What is the most accurate traffic monitoring method?
It depends on the question. For counting vehicles at one junction, video analytics is the most detailed. For measuring how fast traffic is moving across a whole city, probe data is the only method with the coverage to answer at all. Accuracy at a point and accuracy across a network are different problems.
How much does a traffic monitoring system cost?
Camera and sensor systems carry hardware, civil work, power, backhaul, and maintenance costs that scale with every road you add. Software-only systems that use existing probe data have no per-road hardware cost, which is why they can cover a full city for a fraction of a hardware deployment.
Can we use our existing cameras alongside probe data?
Yes, and most cities should. Cameras handle enforcement and junction operations. Probe data handles network-wide congestion and speed. They answer different questions and there is no reason to choose.
TraffiCure delivers real-time traffic intelligence for every road in your city — no cameras, no sensors, no construction. See all features or book a demo to see your city's data.