Smart Traffic System


Image processing system based on deep learning, creating traffic big data, for an effective traffic system (Design of road lines, Calculation of the numbers of lane by direction, and design of signal reality) by extracting traffic information by direction of intersection and vehicle type through traffic object detection and tracking algorithms.


Features & Advantages

Deep learning-based vehicle detection

· Minimize traffic count errors

· Accurate detection in all environments (weather, camera angle, vehicle size and lighting condition)

· Ability to calculate automatically traffic volume by lane

· Ability to quantify chnages in traffic volume
   in each direction

Calculation of traffic volume in all conditions 

(day/night, sun/rain, dry/snow/wet)

System collects deep learning data at various intersections and processes it to achieve an accurate detection rate  despite obstacles (e.g. cables) or weather conditions (day, night, rain, snow, strong winds, etc.)

Vehicle type classification and pedestrian detection

· Classification of vehicles by compact, full-sized and bus to ensure traffic information accuracy

· Ability to count pedestrians walking in each direction

· Object separation capability to achieve accurate count when pedestrians walk as a  group

Accurate traffic count in public or variable lanes and ability to detect U-turns

· Accurate count is achieved by separating the left, center and right sections of a public lane

· Position detection point on vehicles’ travel path in a U-turn lane and utilize vehicle trekking technology to confirm U-turn

· Use schedule-driven algorithm to count variable lanes separately

Smart Traffic Control Program