What is broken-windows?
Broken-windows is an AI created to discover visible crime from a street scene image.
How does it work?
broken-windows has an algorithm that has been trained to recognise visual crime. This allows the AI to identify visible crime from a street scene to highlight areas of crime that may not be identified in traditional methods such as survey and administrative data.
The AI is in its early stages of development and has produced preliminary results for a UK city for where there is data available. There are proposed developments that will increase the accuracy of the AI
By using the AI to identify crime, it is possible to provide a metric on visible crime at a GPS level, potentially in real time. This will allow for targeted interventions to improve well-being and reduce inequalities.
The areas of visible crime identified by the AI have been mapped and displayed by using Leaflet, an open source javascript library.
Please take a look at the blog for more information such as technical information and approaches to data collection.
Further Developments
There are number of improvements that are being made before a full scale version is available.
Make a donation.
If you would like to assist with scaling up the project, please do donate:
Attributions
Data is limited to the images from Mapillary.com. This is an open source street-level image database with contributors using smartphones and action cameras, demonstrating that everyday equipment can be used to collect data on visible crime
Mapillary: CC BY-SA 4.0 Mapillary.com
Portions of this page are modifications based on work created and shared by Google and used according to terms described in the Creative Commons 4.0 Attribution License.
About me
I, John Wildman, am a highly skilled data scientist currently employed by the civil service.
For enquiries and sponsorship please contact me: