Traffic sign recognition using deep learning

Abstract

- Traffic Sign Detection and Recognition can be used as driver assistance which also ensures contributing to safety of drivers, pedestrians and vehicles. The usage personal vehicles cars or Two Wheelers are increased during and after COVID-19. The heavy traffic, congestion and fast driving, leads to traffic accidents which caused a lot of personal injury or property loss. In such scenario, the country like India requires traffic signs on road side so that the drivers aware the status of road and able to convey the important traffic information to the driver. In modern artificial intelligence era, this traffic sign are recognized by intelligence system and can be used for voice or any other applications. In our project, the traffic sign was recognized with highest accuracy using Machine Learning Algorithms. To collect the traffic sign data from the road, the camera from cars are used. The Various factors which affected the identification of traffic signs are lighting factors, light intensity. These factors lead to image exposure, light weakness that result in dim, blurred and corroded image. The Convolution Neural Network deep learning algorithm used to find the accuracy of image recognition. the accuracy Calculated as the value of RMSE or MSE. This project designed an improved CNN uses convolution pooling to extract low-dimensional features and high- dimensional features of images to achieve higher accuracy and lightweight models.