To stop and park vehicles on the roads, special sections are provided to eliminate congestion in the event that one or more vehicles had to stop moving. If the vehicle stops in the wrong place, it can slow down the entire traffic or lead to an accident. In this regard, the identification of stopped vehicles is becoming one of the priorities in video analytics systems.
In conventional video surveillance, the operator does not have the ability to detect all stopped vehicles in a timely manner - he has to analyze too large amounts of data. The neural network copes with this task without problems. The RoadAR system calculates the average position of the vehicle in the image for the last X frames and the average position of the vehicle for X frames Y frames ago. If these positions coincide, and there is no congestion on the lane, then the vehicle is considered to be stopped. Thus, RoadAR reveals:
- facts of intentional stopping of traffic on the carriageway, on motorways, as well as in other places where stopping and parking are prohibited,
stopping the vehicle due to breakdowns and accidents.