Traffic counting: radar or AI video?
There are two tried-and-tested measurement methods available for traffic counting: radar-based detection using the SDR side-looking radar and AI-based video detection using ARGOS. Both methods provide precise, GDPR-compliant traffic data – but they differ fundamentally in terms of detection geometry, classification depth and primary application.
DataCollect provides both technologies from a single source and works with you to select the method that best suits your requirements – from a simple cross-section on an arterial road to a comprehensive junction analysis including turning behaviour.
Two measurement principles for traffic counting
Radar and AI video are not competing systems, but rather two deliberately separatefor different tasks. The following overview describes both principles, before a direct comparison helps you make your decision.
Radar-based traffic counting – SDR side-view radar
The SDR side-view radar is a contactless, self-calibrating traffic counting system. It measures at the road cross-section to the side of the road and records vehicle counts, speeds and vehicle lengths with lane-level accuracy across up to three lanes simultaneously. Configuration is carried out on-site via a smartphone, and the installation height can be freely selected between 1 and 8 metres.
- Self-calibrating – set-up and quality checks carried out directly on-site via an app
- Outputs include speed, vehicle count, object length and freely configurable length classes
- Energy-efficient: mobile battery operation for several weeks, 230 V or solar power for 365 days of continuous operation
- Variants: SDR traffic (bidirectional) and SDR Dual lane (bidirectional and unidirectional)
AI-based video detection – ARGOS
ARGOS is an AI-based video detection system with edge AI: analysis takes place directly within the device (on-board), meaning that analysed traffic data is available without any delay. ARGOS detects, classifies and counts 11 object classes, including pedestrians and bicycles, and covers entire traffic areas – junctions, roundabouts and multi-lane thoroughfares, including cycle and pedestrian traffic.
- Comprehensive coverage of junctions, including turning patterns and source/destination traffic
- 11 semantic object classes, including pedestrians and cyclists
- On-board data processing (Edge-AI) with data transmission via LTE or Wi-Fi to myTrafficData 2
- 100% GDPR-compliant – only anonymised object data is stored
Radar and AI video in direct comparison
| Criterion | SDR side-view radar (radar) | ARGOS (AI video) |
|---|---|---|
| Measurement principle | Radar, contactless | AI video detection (Edge AI) |
| Detection geometry | Road cross-section | Road cross-section, roundabouts, junctions |
| Lanes simultaneously | Up to 3 lanes | Multi-lane + turning relationships |
| Object classes | 4 or 6+ length classes | 11 object classes including pedestrians/cyclists |
| Turning relationships | Not mapped | Fully recorded |
| Subsequent correction | No | Recounting for up to 90 days |
| Installation height | 1–8 m, freely selectable | Pole up to 6 m |
| Measurement operation | Mobile short-term measurement + stationary continuous measurement point | Mobile short-term measurement + stationary continuous measurement point |
| GDPR | Compliant (no personal reference) | Compliant (anonymous object data only) |
| Data platform | myTrafficData | myTrafficData 2 |
Accuracy and data quality
The reliability of the count determines the robustness of the resulting planning. ARGOS’s AI video detection was tested in the Mobility Lab and achieved an accuracy of over 95 per cent – even under heavy traffic conditions with a daily traffic volume (DTV) of between 70,000 and 100,000 vehicles per day.
>95 per cent accuracy in the Mobility Lab test – proven at a daily traffic volume (DTV) of 70,000 to 100,000 vehicles per day.
The radar-based SDR side-view radar delivers consistently high data quality for cross-sectional measurements, with lane-accurate detection of individual vehicles. Both systems are designed for continuous operation and store millions of data records; evaluation is carried out uniformly via the myTrafficData cloud platform.
Classification: How detailed is the distinction?
The most important difference in terms of content lies in the depth of classification – and thus in the question of which analyses will be possible later on.
SDR side-view radar – length classes
The radar classifies vehicles based on their length. SDR traffic distinguishes between four or more freely configurable length classes and is also suitable for installation on bridges. The SDR counts bicycles separately by direction and determines their proportion relative to the rest of the traffic.
ARGOS – semantic object classes
ARGOS distinguishes between 11 semantic object classes and thus recognises not only the size but also the type of road user – from lorries and buses to pedestrians and cyclists. This level of detail is essential for analysing multimodal and complex traffic situations.
GDPR and data protection
Both measurement principles are fully GDPR-compliant. The radar-based SDR does not, as a matter of principle, collect any personally identifiable data. Although ARGOS detects traffic via camera, it stores only anonymised object data – processing takes place directly within the device, and no personal image data is transferred to the cloud. Actual video recording is not necessary with ARGOS.
ARGOS is 100% GDPR-compliant: AI analysis takes place on-board; only anonymised object data is stored and transmitted – no personal video data. No actual video recording is required with ARGOS.
Recounting – adjusting driving relationships retrospectively
One feature that really stands out in comparison is ARGOS’s recounting function. In the cloud, the detection zones and travel relationships can be modified even after the measurement has been completed, and the data can be recalculated – retroactively for up to 90 days.
The recounting function is unique: if the requirements for an analysis change, counting zones can be adjusted retrospectively and the results recalculated – without the need for a new measurement campaign.
When to use radar, when to use AI video?
As a rule of thumb: the simpler the cross-section and the more important continuous operation is, the more likely radar is the better choice. The more complex the junction and the more types of traffic need to be distinguished, the more likely AI video is the better choice.
SDR side-view radar (radar) is the right choice when …
- simple cross-sectional counts are required on one- to three-lane roads
- speed, vehicle volume and length classes are the main focus
- self-sufficient continuous operation over weeks or all year round (solar-powered) is required
- direction-separated cycle counting on cycle paths is needed
ARGOS (AI video) is the right choice if …
- complex junctions with turning relationships or roundabouts requiring source/destination traffic data are to be analysed
- pedestrians, bicycles and other modes of transport need to be recorded separately
- multimodal traffic is to be counted with a high level of classification detail
- analysis areas are to remain adjustable after measurement (recounting)
DataCollect provides both technologies from a single source and, without favouring any particular technology, recommends the method best suited to the specific requirements – without being tied to a single measurement principle.
Frequently Asked Questions
Radar-based counting (SDR side-view radar) measures at the carriageway cross-section and records vehicle volumes, speeds and length classes across up to three lanes. AI video detection (ARGOS) covers entire traffic areas, including junctions and turning patterns or roundabouts, with source/destination traffic detection, and distinguishes between 11 object classes, including pedestrians and bicycles. Radar is suitable for simple cross-sections and continuous operation, AI video for complex junctions and multimodal traffic.
ARGOS achieved an accuracy of over 95 per cent in the Mobility Lab test. This accuracy was also demonstrated under average daily traffic (ADT) conditions of between 70,000 and 100,000 vehicles. The data is analysed directly within the device (Edge AI), meaning reliable data is available without any delay.
Yes, both methods are fully GDPRcompliant. The SDR traffic radar does not collect any personally identifiable data. ARGOS detects traffic via an on-board camera and stores only anonymised object data – no personally identifiable video data is transmitted to the cloud. Actual video recording is not necessary with ARGOS.
The ARGOS AI video detection system, with its 11 semantic object classes, is designed for the differentiated detection of pedestrians and cyclists, as well as other types of traffic. The SDR side radar is also available for counting cyclists in separate directions on cycle paths.
Recounting is a unique feature of the ARGOS AI system: In the myTrafficData cloud, count sections and traffic relationships can be adjusted even after the measurement has been completed, and the data can be recalculated retrospectively for up to 90 days – without the need for a new measurement campaign.
Yes. DataCollect provides both the radar-based SDR side-view radar and the AI-based video detection system ARGOS from a single source, and recommends – without favouring any particular technology – the method best suited to the specific task at hand. Both systems are analysed via the myTrafficData or myTrafficData2 platform.
Further reading: Accuracy of traffic counting systems in the Mobility Lab benchmark and AI video detection and the GDPR.
Are you planning a traffic count and would like to know which method is best suited to your measurement task?