welcome > projects > 66

AI to better urban cleanliness

Publié le 12/01/2022

During the Source of Innovation competition, the company Cortexia won one of five 10,000 CHF checks for improving its waste detection algorithm using Artificial Intelligence.

For many years, Cortexia has been helping cities to monitor urban cleanliness, preserve the environment and save cleaning resources. The Fribourg-based company maps the urban space in real time and shows the level of cleanliness of each street thanks to cameras mounted on sweepers or other technical vehicles. Its technology can collect and detect 40 types of waste, the smallest of which is the cigarette butt. A model created in partnership with the EPFL, trained thanks to numerous images containing waste, collected over the past 4 years.

"In Geneva, we managed to achieve the same level of cleanliness with 20% fewer machine hours: a non-negligible saving in resources and CO2 emissions, not to mention the wear and tear on the equipment," says Pauline Treyvaud, project manager at Cortexia. But today, the company wants to go further.

If the algorithm is efficient during the day, it is much less efficient in low light conditions. However, the road authorities often work at night or in the early morning.

"Artificial intelligence will allow us to recycle our image bank and transform it for each collection scenario using GANs, Generative Adversarial Networks. These create synthetic images from a real image by changing certain parameters, with a high degree of realism."

The Source d'Innovation grant will allow Cortexia to commission a feasibility study from Jean-Philippe Thiran's EPFL Signal Processing Laboratory. If the tools developed prove to be effective, Cortexia will integrate them into its solution to improve the detection and collection performance of cities' waste.

More informations Cortexia website