The developer claims that a whole range of different programs have been created for this, because it can be extremely difficult to determine this. For example, if the camera recorded a car painted in metallic color on a hot sunny day under the bright sun, then it is rather difficult to understand what kind of light is leaving it. Is it reflected sunlight, or is she really having the right lights on?
Here, a specialized neural network comes into play, which, thanks to a cycle of complex programs, determines whether the desired glow is present in the headlights region and checks whether the latter are working at the moment. The creator believes that such work will not be difficult for the program. In addition, he recalled that today there are already cameras that can independently distinguish not only the type of vehicle, but even the model of the machine itself.
Laskin claims that this program records offenses both at night and during the day. However, he clarifies, the footage taken at night will be a more useful evidence base for courts and other similar departments. It is worth considering that, according to him, the main goal of this program is still to prevent these errors at night. When they can be fatal to motorists. While during the day it is more of an educational measure than an emergency.
It is also noted that, although the accuracy of this program is close to one hundred percent (approximately at the level of 85 - 95%), it will definitely not be possible to do without human control. But, thanks to this fact, the drivers have the opportunity to challenge the received fine. True, it is not always possible to do this, and most often this is due either to the unusual shape of the headlights, or to some unexpected weather conditions that have impaired visibility for the camera.
