Abstract
Highway traffic accidents remain the biggest cause of mortality notwithstanding an increase in web traffic awareness. Road accidents pose a serious hazard to people's lives and property in emerging nations. Traffic accidents are caused by a variety of factors, some of which are more important than others in determining how serious an accident is. Data extraction methods can help with the prediction of key aspects of collapse intensity. In this study, utilising Random Forest, it was discovered that a number of characteristics have a strong correlation with the seriousness of highway crashes. Range, temperature, wind chill, humidity, exposure, and wind orientations are the main factors influencing surprise severity. To forecast the severity of traffic accidents, this study blends RFCNN, or Random Forest and Convolutional Semantic Network, with existing deep learning and artificial intelligence models. Comparing the effectiveness of the proposed strategy to a variety of fundamental learner classifiers is necessary. The crash statistics for the United States from February 2016 to June 2020 are among the data considered in the analysis. The RFCNN beat previous models with 0.991 precision, 0.974 accuracy, 0.986 recall, and 0.980 F-score when used to predict the seriousness of accidents using the 20 most important functions.