An AI-solution to detect waste-dumping for Parkius

The customer
Parkius. The pioneers of digital parking management in the Netherlands. At the request of the municipality of Amsterdam, they developed the first digital parking solution in the Netherlands in 2008. End user of the solution: the municipality of The Hague.

The challenge
The Hague wanted to make the city more liveable for its residence. The city was currently scanning images for parking enforcement, but thought the scanning technique mighit improve other city services. Scanned images could show the condition of street furniture, street lighting and the road surface as well as the presence of waste and bulky rubbish. 

Our project focused on the use of scanning vehicles for the detection of so-called fly-tipping, or illegal dumping, next to waste containers.

The solution
The aim of this project is to reduce the nuisance caused by waste-dumping in The Hague by using overview images taken from the scanning vehicles. has built and trained an object recognition model to recognise containers and four different categories of waste-dumping: white goods, bulky waste, boxes/bags and chemical waste. The model was built using computer vision and deep learning.

The containers and the dumped waste detected by the model are merged. Data is stored for each detection, including: location, date and time, serial number of the container, category of the detection and the total amount of dumped items. Using a Tableau-dashboard, the data gives a visual report that is easy to understand, drawn at the district levels.

The result
The system gave trend analyses and pattern recognition that can be used to improve city waste disposals. Over time, the municipality can see where, for example, chemical waste is frequently dumped. Waste collection driving routes can also be altered according to the data models for more efficient response. 

By better monitoring waste-dumping:

  • Illegal dumping can be identified earlier and more accurately and quickly removed, improving the city streets for residents;

  • The cleaning service and enforcement agency are immediately notified in detail and provided the most efficient driving and collection routes; 

  • Cleaning service compliance with service agreements can be measured and adjusted.