Traffic Sign Classifier

Overview

Develop a Deep Leaning Network to classify traffic signs. To accomplish this a Convolutional Neural Network (CNN) will be developed to classify traffic signs. Specifically the CNN will focus on German traffic signs. German Traffic Sign Dataset. Many aspects of TensorFlow, OpenCV, python, numpy, and matplotlib are used to develop the CNN. The CNN code is based on Tensorflow and executed within in a jupyter notebook environment.

Installing and Running the Classifier

The following steps are used to run the pipeline:

1. Install miniconda environment and related packages

   https://conda.io/miniconda.html
   

2. Clone the SDC-TrafficSignClassifier git repository

   $  git clone https://github.com/jfoshea/Traffic-Sign-Classifier.git
   

3. enable cardnd-term1 virtualenv

   $ source activate carnd-term1
   

4. Run the Pipeline

   $ jupyter notebook TrafficSignClassifier.ipynb
   

The random traffic sign images are located in random_traffic_signs directory.

Writeup

A detailed writeup of the classifier and challenges are located here [writeup] (https://github.com/jfoshea/Traffic-Sign-Classifier/blob/master/writeup.md)