Smart Street Scan
Classifying, registering and monitoring street waste via image recognition

Automatic image or object detection is a computer technology that is part of Computer Vision. It is commonly referred to as Image Recognition. Especially thanks to large tech companies, such as Google and Tesla, image recognition is in an advanced stage of use. Take, for example, the rapid developments with self-driving cars. Another important sector employing image recognition is the security sector, where especially facial recognition is used on a large scale.

Image recognition is also used for very specific purposes, such as recognizing the ripeness of agricultural products or distinguishing hand gestures for sign language. In the public sector, on the other hand, and in particular in the management of public space, there are still too few implemented applications. At CentERdata we have experience in the field of image recognition and its applications in the public sector.

Smart Street Scan

Keeping the public space clean is an important core task for municipalities. Advanced technology in the field of Artificial Intelligence (AI) can help. We conducted a feasibility study for the municipality of Utrecht for the implementation of AI techniques for cleaning street waste. This was initiated through an SBIR (Small Business Innovation Research) subsidy from the Netherlands Enterprise Agency (RVO).

With Smart Street Scan we want to automatically scan public spaces using sensor and camera technology. With this, the garbage collection services can be streamlined to a great extent. For example by optimizing sweeping routes. In places where a lot of litter is found, it is also possible to look for ways to change people's behavior. In this way, behavioral science is combined with AI and data science techniques.


The SBIR research has shown that the project is technically feasible in all aspects. Tests were carried out for more than 100 different images and videos, taking into account that street waste never has the ideal shape (think of crushed cans and broken bottles).

Objects were identified and registered with high accuracy during the testing phase. In addition to technical feasibility, specific attention was also paid to organizational feasibility, GDPR compliance, and user-friendliness of the implementation.


In addition to a possible follow-up to Smart Street Scan, we are exploring new ways of research with image recognition with various municipalities and housing corporations.

There are a lot of possibilities with using image recognition in the public sector.

Related projects