TollSense enables operators of highways concessions to automatically detect the right class of each vehicle
passing through the toll gate, only by watching a photo or a video - whatever the classification rules are.
It relies on an expert model based on deep learning and thus improves dramatically the road manager
operational efficiency, by relieving him of most misclassification cases to reprocess.
Already 30 million toll transactions classified
by TollSense at VINCI Autoroutes:
as many transactions to reprocess avoided to operators
To create a high-performance classification algorithm
based on deep learning, cyclope.ai has gathered:
Deep learning gives the algorithm a self-learning ability, which allows it to improve autonomously
Thus, the more data is provided, the more precise the model is and can learn how to recognize
all complex classification cases
Once the performance level required by the operator has been reached, the classification algorithm
can be deployed at scale on the entire road network, through a backoffice integration or directly into the lane.
Now our Tollsense algorithm achieves unequaled
detection performance: trained on already 30 millions of photos,
it takes into account almost all the special cases and recognizes
in just a few milliseconds the vehicle right class.
Tollsense is aproduct easy to integrate, even into complex systems
such as a toll system: it presents itself as a unique API,
easy to connect, and which does not generate any equipment
cost (cameras, civil engineering, etc. .)
Our solution anonymizing and blurring people, silhouettes and plates, make us fully compliant with the new European framework
on the use of personal data (GDPR).
It enables us to create anonymous data sets and thus to store
a large volume of training data, makingitpossible to bring
our algorithms always further.
You are a journalist ?