Why Near‑Miss Data Is the Most Important Metric Your City Isn't Collecting

Every day, thousands of near‑miss events occur at intersections across North America. A pedestrian steps off the curb a moment before a turning vehicle passes. A cyclist swerves to avoid a door opening. A driver brakes hard to avoid a red‑light runner. These moments happen constantly — and almost none of them are recorded.

Near‑miss events are the strongest predictor of future accidents. They are the leading indicators of intersection safety. Yet most cities collect no data on them at all. The behavioral data that could prevent accidents is invisible to the systems that manage our roads.

Here's why that needs to change — and how Canny Vision is making near‑miss data available to cities for the first time.

What is a near‑miss event?

A near‑miss is a situation where a collision almost occurs. At intersections, near‑misses manifest as evasive maneuvers — hard braking, sudden swerves, close passes between vehicles, pedestrians, and cyclists. They are the behavioral signature of unsafe conditions before an accident happens.

Unlike collisions, near‑misses don't produce incident reports. They don't generate insurance claims. They don't appear in police databases. They happen, and then they vanish — leaving behind no trace for the engineers and planners whose job is to prevent accidents.

Why incident reports are not enough

Traffic safety programs have traditionally relied on incident reports to identify dangerous intersections. A high number of collisions triggers a safety review. This is reactive — the accidents have already happened, and in some cases, lives have been lost.

Research consistently shows that near‑miss events occur hundreds or even thousands of times more frequently than collisions at the same location. By the time a collision analysis is triggered, the intersection has already accumulated a long history of near‑misses — events that, if detected, could have prompted corrective action earlier.

The Toronto Vision Zero Road Safety Plan has set ambitious targets for eliminating traffic fatalities. Meeting those targets requires a shift from reactive to proactive safety management. That shift depends on near‑miss data.

How behavioral detection changes the equation

Canny Vision's edge‑based detection platform was built specifically to address this gap. By processing video directly on the camera at intersections, our technology identifies behavioral patterns — including near‑miss events — in real time.

When a near‑miss event occurs, it is logged, categorized, and timestamped. Over time, these events aggregate into a dataset that reveals which approaches are most dangerous, what times of day see the most risk, and which types of conflicts are most common — pedestrian‑vehicle, vehicle‑bicycle, vehicle‑vehicle, etc.

This data gives traffic engineers and safety officers the evidence base they need to make proactive decisions. Instead of waiting for collisions to happen, they can intervene based on leading indicators.

Near‑miss data and Vision Zero

Vision Zero is a strategy to eliminate all traffic fatalities and severe injuries, while increasing safe, healthy, equitable mobility for all. The principle is that traffic deaths are preventable, and that systems design — not individual behavior — is the primary lever for change.

To design safer systems, cities need to understand where, when, and why unsafe interactions occur. Near‑miss data provides that understanding. It is the feedback loop that Vision Zero requires but has historically lacked. Canny Vision closes that loop.

With near‑miss data, cities can:

• Prioritize safety investments based on actual risk
• Measure the effectiveness of safety interventions
• Demonstrate progress to stakeholders and the public
• Identify locations that need attention before they produce collisions

Toronto: a city that understands the stakes

Canny Vision was founded in Toronto — a city with one of the highest rates of pedestrian‑involved collisions in North America. Toronto's Vision Zero Road Safety Plan is among the most ambitious in the region, aiming for zero traffic fatalities by 2031. But without near‑miss data, the city's safety teams have been working with incomplete information.

Canny Vision is deployed in Toronto and Ottawa, providing the near‑miss data that enables proactive safety management. These deployments are proof that the technology works and that the data cities need is finally within reach.

The path forward: near‑miss data as a standard metric

Just as cities track vehicle counts, travel times, and collision statistics, near‑miss data should become a standard metric for intersection performance. It is the closest thing we have to a leading indicator of safety outcomes. Without it, cities are operating blind.

Canny Vision makes near‑miss data available at scale, on existing camera infrastructure, at a fraction of the cost of traditional traffic studies. The technology is here, the data is waiting, and the cities that adopt it will be the ones that lead the next generation of traffic safety.