Traffic-Sign-Classification

This project is submitted towards completing Self Driving Car Nanodegree from Udacity. In this project, we develop a LeNet based neural network that detects the traffic signs commonly observed by a car on roads. A German dataset of traffic sign images that is considered as benchmark is used for training and validating the network. The database contains more than 30000 images of various traffic signals. The images were captured in different situations. The architecture of the network is inspired from one of the most celebrated researcher Yann LeCunn who pioneed the LeNet that was actually made for handwritting detection. The accuracy of the network is around 96%.