NovelSense is developing a battery powered and camera based traffic counter under the ABAKUS brand. The small build-size and low-power design increases the flexibility of the sensor and makes it ideal for temporary installations. Currently the ABAKUS is optimized for the recognition of bicycles. This makes it the perfect tool for cities or traffic planners working on sustainable mobility concepts.
The project “E-Scooter Det” is done in cooperation between the TECO Institute at the Karlsruhe Institute of Technology (KIT), the Smart Data Innovation Lab (SDIL) and NovelSense. Goal of the project is the reliable detection of E-Scooters using the ABAKUS, while at the same time improving the general performance and accuracy of the current system.
I recorded the following video as a contribution to an SDIL online conference.
The ABAKUS system recognizes objects on the basis of camera data. This data is computed using a low-power AI processor running a Neural Net. The recognition pipeline allows the classification and detection on real-time video data. Simultaneously tracking of the recognized objects allows for a detection of direction and speed for each object without the need to store video data.
NovelSense hopes to be the first company to reliably detecting E-Scooters with an embedded neural net. This will open opportunities for multiple use-cases, including a GDPR compliante traffic counter that can reliably differenciate E-Scooters from other traffic participants. This can for example be used to generate reliable traffic data for the growing number of Scooters on European streets.
More information about the project can be found on the SDIL website.