Convolutional Neural Network with Keras with Improved Accuracy and Comparison to Logistic Regression for Traffic Sign Recognition

Abstract

AIM: To predict the accuracy of traffic sign recognition using a Novel Convolutional Neural Network over a Logistic Regression. Materials and Methods: At different events, CNN with test size (N=10) and  Logistic Regression with test size (N=10) the sigmoid limit is a probability assumption used in basic CNN that aids in chipping away at the figure of accuracy.  Logistic regression and CNN attained relevance there was a real significance among logistic regression SPSS statistical analysis yielded an accuracy value of p=0.001(2 tailed) (P<0.05).Result: The findings showed that CNN achieved fundamental outcomes with 91% precision and outperformed  logistic regression with 80%  exactness. Conclusion: CNN is the simplest and most effective algorithm for assembling quick AI models.