The challenge
Graffiti often appears overnight on city walls. For municipalities, finding each new spot and organising its removal is difficult and time-consuming.
Manual detection is inefficient and delays removal. The municipality needed an automated, precise and efficient way to detect and locate graffiti across the city quickly.
Our solution
We developed a model for real-time detection and localisation of graffiti. A vehicle fitted with a 360-degree camera patrols the city streets, and deep learning algorithms detect graffiti with high accuracy and photograph it as the vehicle passes. Each detection is tagged with the GPS location of the vehicle at that moment.
The detections are transferred to an interactive map. Each marker shows the exact location of a graffiti spot together with its 360-degree image, and the platform extracts summary metrics for the whole city.
- Real-time graffiti detection from a vehicle-mounted 360-degree camera
- GPS location attached to every detection
- Interactive map with a marker and 360-degree image for each graffiti spot
- Metrics such as total number of graffiti spots, graffiti density map and average graffiti size
- Adaptable to detect other objects, such as trees, garbage and traffic signs
The outcome
By automating the detection and localisation of graffiti, the platform saves the municipality substantial time and resources. City cleaning teams can identify and prioritise removal quickly, allocate resources better and keep streets cleaner.
Because the model can be adapted to other object types, municipalities can use the same platform for tree maintenance, waste management and traffic sign monitoring, supporting both safety and the appearance of the city.
