CIVIC
Community-centred Infrastructures for Visible and Inclusive Cities — Neighbourhood Evidence Observation Network
A network of AI-assisted traffic sensors deployed in a residential London, Ontario neighbourhood to measure pedestrian, cyclist, and motor vehicle activity.
The problem
Municipal pedestrian and cyclist counts are concentrated in downtowns and on commuter corridors. The places where the data is thinnest — bus stops outside schools, sidewalks in lower-income neighbourhoods, corners where crashes keep happening — are precisely the places where evidence matters most for capital decisions. When a street is invisible in the data, it stays invisible in the budget.
What we built
CIVIC is a network of AI-assisted sensors that count people walking, cycling, and driving at 16 sites across a residential neighbourhood in London. Each sensor runs a small vision model on the device itself. It produces counts, headings, and modal classifications — and nothing else. No video leaves the sensor. No image is ever stored.
How sites were chosen
Sites were not picked by traffic engineers. They were picked through a community siting process with residents, neighbourhood associations, and a youth advisory group. Residents were asked a simple question: "Where do you wish someone was counting?"
What we're learning
Early data is already overturning assumptions. A crossing the city classified as low-priority is in fact the second-busiest pedestrian site in our network during school hours. We're publishing findings as they stabilize — see the Insights section.
Principles
Four commitments we won't compromise.
01
Privacy by design
All vision processing happens on the sensor. No images, video, or identifying data ever leave the device. Cities get counts, not surveillance.
02
Community siting
Residents choose where the sensors go. The data agenda is set by the people whose streets are being studied.
03
Open methods
Hardware specs, model details, and validation protocols are published openly so the work can be replicated and challenged.
04
Useful outputs
Every dataset is paired with a plain-language brief written for the people who need to act on it — councillors, planners, residents, journalists.
Want a sensor on your street?
We're expanding the network through partnerships with municipalities and community organizations. Let's talk.
Contact the lab →