Development of a front-end part of a system that uses a neural network to track traffic on the streets, to analyze traffic and parking congestion and traffic violations, and helps to optimize urban infrastructure.
Our team is developing a front-end part based on React, Redux, Mapbox GLjs, d3.js.
The system constantly takes photos of all objects with a specified time interval and initiates video recording for the certain events involving vehicles.
Everything that happens to objects in the camera's field of view is recorded, and the database stores all the information recorded and processed by the system. If necessary, it can be filtered and downloaded for use.
The system is based on deep learning. Due to this, if at the start of the project the neural network was trained by people by manually marking up fixed objects, now operators can only occasionally get involved in the process, helping the system to minimize errors.