Photo by erin m
D.C. transportation officials are asking the public to help train computers to recognize where crashes are likely happen in order to prevent traffic-related deaths and injuries in the future.
Mayor Muriel Bowser and D.C. Department of Transportation officials announced the Video Analytics Towards Vision Zero project—a partnership between the District and Microsoft that will use videos of near-miss collisions to make roads safer for drivers, passengers, cyclists, and pedestrians.
The project is part of D.C.’s Vision Zero plan to eliminate traffic-related fatalities and injuries by 2024. This year, there have been 23 traffic fatalities in D.C. compared to 17 this time last year, according to D.C. police records.
Traditionally, transit officials have used crash data to help them identify intersections that need to be improved for safety. They say that this new technology, with the help of residents, will identify problems before crashes happen.
“Using video analytics to achieve Vision Zero is one more way we are building a smarter, safer, stronger D.C,” said Mayor Muriel Bowser, in a release. “Residents know traffic issues in their neighborhoods better than anyone, and now we will be able to leverage their knowledge with our existing camera infrastructure in order to prevent crashes and injuries before they occur.”
There are currently more than 130 cameras around the city that record traffic conditions at intersections. Microsoft has created a way for D.C. residents to identify objects in those recordings in order to teach the technology how to distinguish between different movements and types of transportation.
For instance, a person would be able to view the videos and identify a car that’s driving along a road or a pedestrian who is walking across a street. Ultimately, computer algorithms will be able to analyze millions of hours of footage to help predict hazards.
Dr. Victor Bahl, director of mobility and networking at Microsoft Research, explains it this way: “If a bicyclist and a car are on a collision course and at the very last minute, the bicyclist realizes that it can stop and the driver stops, that’s a near miss. Now, if we feed this data to a computer and it starts to recognize a pattern and we feed it many such video frames, it will now know what a near miss looks like.”
The more videos that are labeled by residents, “the better the computer system will become and effectively it will become much better at saving lives,” Bahl said.
Here’s a video on how the labeling process works:
