Fixed Roadside Installations
Deployed on roadside or gantry-mounted cameras, enabling continuous vehicle categorization across one or multiple lanes.
For tolling operators, bridge, tunnel and port operators, enforcement agencies and traffic authorities, distinguishing between a passenger car and a truck is only the beginning.
Charging schemes, compliance checks, and traffic analytics often depend on detailed vehicle information, including axle count, trailer configuration, and vehicle class. Obtaining this data accurately and consistently at traffic speed remains a significant operational challenge.
Asura's AI-powered Vehicle Categorization solution automatically categorizes vehicles using standard camera infrastructure, delivering detailed vehicle metadata in real time without the need for roadside sensors or manual inspection.
Field measurements conducted by a European national toll operator on live toll-road traffic, across multiple sites and at different times of day, demonstrated the following combined results:
Accuracy counts only vehicles and plates that are visually detectable and recognizable in the recording camera's image, and compares correctly recognized data against them. Source: customer reference letter.
The platform uses advanced deep-learning models to analyze vehicle geometry, axle configuration, and trailer presence directly from video streams.
Unlike conventional categorization systems that rely on a limited set of categories, Asura distinguishes 25+ detection categories, combining vehicle type, axle count, and trailer configuration.
The system also identifies trailers and semi-trailers as separate objects and incorporates them into the final vehicle categorization.
These categories can then be mapped to any national tolling scheme, regulatory framework, or operator-specific categorization model without retraining the AI engine.
Many tolling and enforcement applications depend not only on vehicle type but also on the exact number of axles in contact with the road.
Asura detects and evaluates each axle individually, including the ability to distinguish lifted axles from load-bearing axles. This prevents incorrect vehicle categorization and helps avoid charging or enforcement decisions based on inaccurate axle counts.
Vehicle categorization must remain reliable throughout the day, regardless of environmental conditions.
Asura's AI models are designed to operate across:
This supports consistent categorization in environments where commercial traffic operates around the clock.
The solution supports multiple deployment scenarios:
Deployed on roadside or gantry-mounted cameras, enabling continuous vehicle categorization across one or multiple lanes.
Integrated into enforcement vehicles, allowing authorities to perform vehicle categorization without installing dedicated roadside infrastructure.