Multiclass Classification of Road Traffic Signs Using Machine Learning Algorithms

Modern world’s vehicles are fully simplified and intelligent by means of automatic driving and the detection of road traffic signs is very much critical. While driving, the driver sometimes accidently fails to pay attention of the road traffic signs which are present on the streets. This can be dangerous for the driving person and also for the individuals with him. With an effectual Smart Driving Support System (SDSS) the driver can be informed about the signs. Detection and recognition are the two stages of this smart system based on its features and meaning. The phase of detection is achieved by Canny Edge Detector. The recognition stage is achieved by implementing Convolutional Neural Network (CNN) due to its impressive feature and performance. Further, we also implement other classifiers such as Support Vector Machine, K-Nearest Neighbor, Random Forest and Extreme Gradient Boosting. These models are implemented to check if their performance is better than CNN. The highest accuracy of 97.3 percent was obtained for Convolutional Neural Network and the lowest of 49.08 percent for SVM (Support Vector Machine). The standard GTSRB data set with multi classes is used to show the efficiency of the proposed approaches. Keywords: Smart Driving Support System, CNN, K-Nearest Neighbor, Random Forest, SVM, Extreme Gradient Boosting