After three and a half years, our successful cooperation project under the mFund funding programme came to an end on 30 June 2026. Together with the City of Leipzig, Leipzig University, Stadtreinigung Leipzig (the city’s waste and street-cleaning service), Aufbauwerk Leipzig and ccc Software GmbH, we drove the “Digital Design of Leipzig’s Urban Space” (DiGuRaL) forward significantly.
We thank all partners for their constructive collaboration under demanding conditions and look forward to feeding the extensive project results directly into our Cyface solutions.
The focus of the project was to equip the waste collection vehicles of Stadtreinigung Leipzig with our data collection app. During regular operation, image data was recorded to gain valuable information about the urban space. This approach has the advantage that new data is continuously recorded and old data updated, because the waste collection vehicles are out in the city every day. This eases the workload of municipal staff, such as the road maintenance depot, in meeting its statutory duty to keep roads safe.
The project was divided into three central building blocks:
The highest data-protection standards in the municipality
Working with public bodies places particularly high demands on data protection. Based on an extensive data-protection assessment by the City of Leipzig, we precisely adapted our app to these requirements. In doing so, modern anonymisation techniques developed by Leipzig University were integrated directly.
With our many years of experience in data security, we were able to actively help shape essential parts of the data-protection assessment.
Efficient image data collection during regular operation
We supported Stadtreinigung Leipzig in equipping the large waste collection vehicles with suitable smartphones for data collection. Thanks to the openness of the staff and the final approval by the works council, we were able to continuously record and analyse image data over two months without disrupting the daily workflow.
An AI model for automated road segmentation and damage detection
Using the recorded data and a categorisation provided by the City of Leipzig’s civil engineering office, we developed an innovative AI model for segmenting road areas. This allows us to focus the automated detection of road damage precisely on the relevant carriageway area.
The model was trained with 1,800 images from practice, annotated by hand by Cyface GmbH staff. This already produced usable results. A future expansion of the dataset is, however, planned.
Advance the digitalisation of your transport infrastructure!
The insights and AI models we gained will feed into our products over the coming months. Would your municipality or company like to carry out a similar project? We are happy to support you with our technology and expertise.
Get in touch
