CertAIn - AI for improved waste flow statistics and higher sorting rates at recylingcenters

Important results from the project

We have achieved the goals, and the project has demonstrated a proof of concept for how AI-based computer vision models can be a tool for statistics and analysis in waste management. The project has enabled municipalities to test new methods in waste sorting and establish new collaborations for their innovation activities. There is still a need for further development, and the project has created a strategy to go from proof of concept to a solution ready for implementation at recycling centers.

Expected long term effects

There is strong interest in continuing the work, and the municipalities see high potential for follow-up projects to result in new approaches and development strategies for higher sorting rates, based on actual data and well-informed decisions. In the long term, the project will have contributed to supporting circular material flows and achieving the municipalities’ ambitious environmental goals.

Approach and implementation

The project has established methodologies for automated data collection and for training AI models using information derived from real sorting processes. Datasets were created from camera observations of 10 different waste categories at recycling centers in Gothenburg, Kungsbacka, and Alingsås. It has been demonstrated how computer vision and automated systems can reduce the need for manual annotation and support the development of robust classification models in municipal waste management.