In this project, I apply my learning of neural networks and convolutional neural networks to classify traffic signs. I train a model so it can decode traffic signs from natural images by using the German Traffic Sign Dataset. After the model is trained, I then test my model on new images of traffic signs I found on the web.
I developed my model on the basis of the LeNet-5 architecure, applying various types of tuning in the process of model training. The data set consisted of over 50,000 images. I trained the model on Amazon EC2 Medium instance, which took approximately 1 minute of training per epoch.
The detailed steps, code and discussion are included in the Traffic_Sign_Classifier.ipynb notebook included in this repository.