Detecting and interfacing urban litter using computer vision.

A data structure is developed to facilitate on-edge litter detection and the methodologies and techniques to interface the data are discussed. Data on litter occurrences have been modeled according to the essence of their use-case, grouped to road sections. The system has been developed to serve both users (contractors, maintainers and managers at the municipality) as well as external software to optimize sweeper routes. The viability of the system has been proven and considerations and advice are reflected upon to allow future development and improvement.