Smart Computer That Learns from Videos Can Make Traffic More Efficient

traffic monitoring image

Image credit D. Küttel / ETH Zurich; via PhysOrg

ETH researchers Daniel Küttel and Michael Breitenstein teamed up with professors Luc Van Gool and Vittorio Ferrari from the Institute of Image Processing to create a new software program that can learn from watching moving objects, analyzing things like street scenes and figuring out patterns and habits of things like moving vehicles. The new technology allows the computer to recognize things like the movements of normal traffic flow and any changes in that "normal" situation. It may have a big use in analyzing traffic and improving flows in congested areas, or running traffic lights to make intersections safer. According to ETH Life, "To create this programme, the researchers mounted cameras at a number of junctions around the city of Zurich and recorded hours of video clips. The computer then analysed these video sequences and automatically, i.e. without intervention by the programmers, established rules governing the flow of traffic. The computer had to spend about a day working out the calculations for each hour of video footage that had been recorded. Once the machine had 'learned' the standard patterns, however, it was then able to interpret the video recordings in real time."

When the researchers compared the computer's work to tram timetables from that day, they saw the computer was extremely accurate and the computer was analyzing the scenes it saw correctly.

The software still needs to find a home in traffic coordination, but the potential is there for everything from notifying traffic control centers when a jam or accident happens to regulating the flow of lights.

Follow Jaymi on Twitter for more stories like this
More on Computers Helping Traffic
Microsoft's Predestination Makes Your Car Like Knight Rider (Video)
Networked Traffic Lights Could Save Time, Fuel, and Lives
Diagonal Crosswalks in L.A. to Make City More Pedestrian-Friendly

Related Content on