Improving the transfer learning performances in the classification of the automotive traffic roads signs

This paper represents a study for the realization of system based on Artificial Intelligence, which allows recognition traffic road signs in an intelligent way, and also demonstrates performance Transfer Learning object classification general. When systems are trained aspects human visualization (HVS), helps or generates same decisions, construct robust efficient systems. us to avoid many environmental risks, both weather conditions, such as cloudy rainy that causes obscured vision signs, but main objective is all risks dangerous achieve safety, accidents due non-compliance with rules, vehicles passengers. However, simply collecting different places does not solve problem, classifying needed improve safety people its environment. proposed sign extracts visual characteristics from Convolution Neural Network (CNN) model. model aims assign class image through classifier most optimized. Then evaluation effectiveness according several criteria, using Confusion Matrix report, in-depth analysis results obtained by images taken urban world. The encouraging comparison developed scientific literature, example, Advanced Driving Assistance Systems (ADAS) sector automobile.