Recognition and Traffic Warning of Severe Weather using Machine Vision

Severe weather has a significant impact on road traffic operations, safety, and autonomous driving, making it susceptible to serious accidents due to drivers' erroneous perceptions of the environment and inappropriate driving speeds. To prevent and reduce traffic accidents in severe weather conditions, providing accurate speed warning information is essential for ensuring the safe operation of motor vehicles on highways under adverse weather conditions. Firstly, based on severe weather image data, different levels of severe weather are categorized according to traffic meteorological condition classification standards. Through the integration of deep neural network theories, an automatic monitoring model is established to identify the captured video frame information from highway surveillance. Secondly, based on the driving safety speed model, visibility, rainfall intensity, road surface friction coefficient, and other indicators are combined to design a safety speed limit warning system. Finally, a severe weather recognition and warning system along with a visualization tool are designed and deployed on the Web platform. The practical results demonstrate that the Resnet50 recognition model based on severe weather images exhibits good data adaptability and accurate identification results in sunny, rainy, foggy, and mixed datasets.