AI solution to prevent traffic accidents by detecting lane violations, a risk factor in road traffic
Outline
In 2021, the NIA's artificial intelligence learning data rescue project successfully completed the "Lane Violation Data Using Lane Violation Image Data" task, and based on this, the development of a real-time Lane Violation AI solution was completed.
Process
Detection
Vehicle Detection Algorithm
Lane Detection Model
Class of Lane
Item
Detail
Lane Type
Solid Line / Dashed Line / Double (Solid-Dashed) / Double Solid Line
Solid
Single, Double
Dashed
Single, Double
Combined
Left Dashed + Right Solid / Left Solid+Right Dashed
Lane Color
White/ Yellow / Blue
Traffice Violation Detection Algorithm
Traffice Violation Classification
designated lane violation
designated lane violation recognition
Recognition of driving on the primary lane of large trucks, etc.
Illegal lane change
solid lane change recognition
center line violation recognition
Bus-only lane violation
Bus-only lane recognition
Blue double solid line + center line (yellow double solid line)
Blue double solid line + road end
Recognition of non-bus vehicles in exclusive bus lane zones
Recognition of center line or end lane
Blue Lane Recognition
Based on the blue lane, the lane closest to center line or end lane is the bus-only lane.
Can be used in various ways such as vehicle recognition, lane recognition, traffic law violation, etc.