Accidents detection

In conditions of heavy traffic, any accident becomes a cause of congestion. In most cases, the congestion persists for several hours, which is often due to the untimely adoption of measures to eliminate the consequences of an accident. The RoadAR system allows you to quickly detect the facts of an accident and notifies the operator of the need to act according to the instructions.

Solution metrics

99%

Recognition accuracy

Accuracy measured on test videos in the daytime
Up to 1/day

False positive

Number of false positives per day per camera
10 FPS

FPS

Frames per second
FullHD

Video quality

Recommended video stream quality

Technology Description

A traffic accident is not uncommon in a community with heavy traffic. Regardless of how many vehicles are involved in an accident, an accident almost always results in congestion or at least traffic obstruction. And here the main thing is to take action in time.

RoadAR video analytics recognizes and captures various types of accidents, creating an appropriate event with operator notification. The neural network detects not only the collision of vehicles with each other, but also collisions with road infrastructure objects, cars falling into a ditch, etc.

High recognition accuracy ensures prompt detection of almost 100% of accidents.
A self-learning neural network uses the accumulated data sets to improve the detection efficiency.