C.L. Dai, J. Wang

Pages: 77-86

Abstract
In order to overcome the poor accuracy of traditional road traffic safety risk assessment methods, a fuzzy road traffic safety risk assessment based on data-driven technology is proposed.Firstly, the attributes of road traffic risk events and regions are described based on data-driven technology, and the risk assessment model is described; Secondly, the risk probability of road section is calculated by using the semantic information of scene context; The state transition coefficient in DBN network is obtained through data driving, and the empirical data of traffic safety risk is allocated to realize road traffic safety risk assessment.The experimental results show that the accuracy, recall, F1 score and accuracy of traffic safety risk assessment of this method are 0.9357, 0.9248, 0.9215 and 91.43% respectively, indicating that this method has good performance in predicting the impact degree of traffic accidents.
Keywords: genie software; dbn network structure; data driven; security scenario; including semantic segmentatio