Cameras are already everywhere on our roads. Computer vision turns their video into traffic data: it detects and classifies every vehicle and pedestrian, follows their paths and measures how the traffic behaves, without counting by hand. It is one of the building blocks of an Intelligent Transportation System.
From video to traffic data
At Think Transportation we built our own computer vision application, Vision Genius. Its video analytics pipeline locates and categorises vehicles and pedestrians in traffic videos, tracks their movements and extracts parameters such as total and interval counts, lane counts, traffic flow and density (see Automated Extraction of Micro Traffic Parameters).
What it makes possible
- Vehicle and pedestrian detection and tracking: the precise, real-time data that adaptive signal control, ramp metering, signal synchronisation and speed advisory depend on.
- Pedestrian behaviour analysis: posture estimation from key points on the body, with uses from autonomous vehicles to safety and surveillance.
- Hot spot and congestion identification: where people and vehicles concentrate, on road corridors as well as in shopping malls, showrooms and service centres.
- Smart toll management and intelligent parking: automatic vehicle identification and classification at toll plazas and parking facilities.
Why it matters
Manual traffic counts are slow and costly, and they cover only a few locations for a few hours. Video analytics can cover more locations for longer and repeat the analysis whenever it is needed, so the traffic studies and designs built on the data rest on a fuller picture. Together with our traffic survey app and real-time geospatial data collection app, it brings field data into the office faster.
Want to turn your camera footage into traffic data? Contact our team.





