---
lang: en
title: "AI traffic monitoring and vehicle counting"
url: https://nabrio.com/use-cases/traffic-monitoring
canonical: https://nabrio.com/use-cases/traffic-monitoring
updated: 2026-10-27
publisher: Nabrio Co., Ltd. (nabrio.com)
---

# AI traffic monitoring and vehicle counting

Nara turns roadside cameras into traffic sensors. It counts vehicles per lane and direction, classifies cars, motorbikes, buses and trucks, measures density and queue length, and alerts the command centre to congestion, all processed at the edge so video does not have to be streamed to a central cloud.

- Sectors: [Smart city](https://nabrio.com/industries/smart-city.md), [More sectors](https://nabrio.com/industries/more-sectors.md)
- What it takes: Ready to configure. Works with Nara’s built-in models. You set up the zones, rules and outputs. Accident detection uses a model trained in Reva or the AI Decision node.
- Products: [Nara](https://nabrio.com/nara.md)

## The problem

Cities need live traffic data and early warning of congestion and incidents without streaming every camera to a central cloud.

## How it works

- **Input:** Roadside and junction cameras
- **Process:** Vehicle detection and tracking: counts per lane and direction, classification, density, queue length
- **Action:** Dashboards and congestion alerts to the command centre over MQTT or HTTP

Nara detects and tracks vehicles, counts them per lane and direction with lines and zones, and classifies each one as a car, motorbike, bus or truck. Density and queue length come from zone counts and dwell time. Results and congestion alerts go to the command centre over MQTT or HTTP, and video stays at the edge.

## Variations

### Traffic flow and congestion

Vehicle detection and tracking give counts per lane and direction, density, queue length and congestion alerts.

### Vehicle classification counts

Cars, motorbikes, buses and trucks counted per hour.

### Accident detection

Stopped vehicles in live lanes, collisions and crowds forming raise an alert to the command centre. This uses a model trained in Reva or the AI Decision node.

## What you need

- Roadside or junction cameras
- Edge devices (NVIDIA Jetson) or on-site servers running Nara
- A command centre system reachable over MQTT or HTTP

## Nara building blocks

- Camera nodes
- Object detection (vehicles)
- Tracker with line and zone counts
- Dwell time
- Dashboard
- MQTT and HTTP to the command centre

AI models: [General object detection](https://nabrio.com/nara/ai-models.md), [Object tracking and analytics](https://nabrio.com/nara/ai-models.md)

## What it improves

- Live traffic data
- Faster congestion response
- Bandwidth saved by processing at the edge

## Frequently asked questions

### Does video have to be streamed to a central server?

No. Nara processes video at the edge, on NVIDIA Jetson devices or on-site servers, and sends only results to the command centre over MQTT or HTTP.

## Related use cases

- [Traffic violation detection](https://nabrio.com/use-cases/traffic-violation-detection.md)
- [Licence plate recognition](https://nabrio.com/use-cases/licence-plate-recognition.md)
- [People counting](https://nabrio.com/use-cases/people-counting.md)

Talk to us about traffic monitoring: sales@nabrio.com. [Download Nara](https://nabrio.com/download.md).