Winner of the 'Best Public Sector Project' at the 5th National Technology Awards in October 2021, and voted by the public as a top 3 finalist at the DigiLeaders 100 'AI Innovation of the year' 2020, the proof of concept began for Swindon Borough Council with a conversation about whether proactive identification could address every day issues such as potholes, graffiti and fly-tipping?
Swindon Borough Council’s Emerging Technology team was set up in 2019 with the objective of using new technologies and innovative ways of working to design better public services.
Solution
One of their first projects, in collaboration with Methods, developed a ground-breaking Proof of Concept that uses Object Detection and Machine Learning to enable streamlined and channel-agnostic reporting of street issues, combined with advanced data visualisation to predict issues and expedite fulfilment.
Swindon built on this PoC to deliver the ‘Report It’ project, in response to the challenge of dealing with over 300+ incidents of fly-tipping reports per month. In many of these cases, the accuracy of the location, the type, and size of the discarded items would be unknown.
The solution was to use open architecture where possible, alongside the Council’s existing Jadu forms and case management solution. Amazon Web Services (AWS) was used to apply the Rekognition cognitive service, enabling an understanding of what items would be within an image using its image recognition and Machine Learning algorithms, as well as open data sources from around the web to provide source dimension and weight data.
Finally, open source and free tier tools provided route optimisation and prioritisation of the daily jobs list to the Streetsmart team via the Jadu case management solution.
Impact
This real-time pothole detection, that can be fixed upon any of the councils vehicles e.g. refuse collection lorries, and collects the data simultaneously. This technology can also be used to identify and prioritise graffiti removal as well as locating fly-tipping.