Аннотации:
The need for remote road traffic monitoring is essential to reduce accident possibilities.
It encourages drivers to abide by the traffic laws. Recently, researchers have been
focused on automatic traffic monitoring using edge devices, Computer Vision (CV),
and Machine Learning (ML) due to the increase in vehicle numbers. However, due
to the number of edge devices, the problems of high-cost hardware for cloud-based
computations and scalability of the bandwidth are arising. Therefore, this paper
proposes a low-cost microprocessor-based traffic monitoring system that will conduct
all processing on the edge. The system will be used near traffic lights and detect
law violations on red lights. It will make drivers more careful by adding certain
consequences which will lead to fewer accidents. The microprocessor is equipped with
a camera module and is used to run a video processing algorithm and Convolutional
Neural Network (CNN) for law violation detection, and its further classification. The
device will be installed on the traffic light pole in Astana, Kazakhstan