Tollsense is an AI-powered image processing software, able to detect in real time the class of a vehicle at the toll plaza, from a unique image feed and with a very high level of accuracy
Despite the high number of equipment already set up at the toll plaza (optical radars and other lasers), 5 to 10% of toll transactions are still commonly manually reprocessed by operators: a tedious and time-consuming process, expensive for tolling operators and generating numerous errors.
Based on the latest Deep Learning technologies and trained on 101 million images, Tollsense automatically detects the class of vehicles at toll booths with the same performance as a classification expert.
By eliminating the manual reprocessing of classification errors, operators immediately
gain in productivity: great impact on toll function ROI.
Thanks to the reduction in intercom customer calls, toll lanes are more fluid and customers more satisfied (decrease in customer claims).
Ready to scale
Tollsense is already deployed at an industrial level throughout the VINCI Autoroutes network.
It is the only AI product to have been deployed at scale in France in 2018.
With our anonymization technology, we respect user confidentiality by blurring personal data visible on the images we process (GDPR compliant).
Management and analysis tools for supervision, easy to learn by operators (simple and intuitive interface)
A confidence rate is assigned to each response, that can be used as a filter to ensure a specific level of reliability
The API responds in less than 250 ms (cloud or on-premise processing), allowing the solution to be integrated directly in the toll IT system
A single API, easy to integrate into your toll systems: directly into the lane software or in your back-office, without any equipment costs
4 443 km network
320 toll plazas
880 M transactions / year
VINCI Autoroutes Toll Division
2018 - 2019
54 300 hours per year of manual reprocessing
35% of a toll operator’s working time
After only 4 months of R&D, Tollsense was deployed at scale throughout the network for all class 2 and 5 checks, directly into the business qualification tool.
of the manual reprocessing
for toll operators
Our Tollsense algorithm has reached unprecedented performance levels: trained on over 101 million images, it integrates almost all rare cases. It can easily be adapted according the specific needs of each client, to reach a tremendous performance:
Deep Learning enables the algorithm to self-learn and continuously improve. Thus, the more data is provided, the more accurate the model is and the more accustomed it is to recognizing complex cases such as snow, trucks, trailers, etc.TEST ON MY DATASET
Once the solution reaches the required performance level, the classification algorithm can then be deployed in production on the network. This can be done via integration in back office or directly on the toll lane.TEST ON MY DATASET
Real-time vehicles classification via video analysis, for a better knowledge of your traffic
Revolutionizing tolling by providing a camera-based-only Free Flow gantry, at a lower cost