ENSC 813 (Deep Neural Networks) Project - Traffic Sign Recognition Convolutional Neural Network

The goal of the project was to design a convolution neural network (CNN) using TensorFlow for computer vision. For this project, the CNN was designed to identify six different traffic signs-

1. Crosswork
2. Hazard
3. School Zone
4. Stop Sign
5. Turn Lane
6. Yield

The following aspects of the CNN needed to be adjusted to yield the most accurate results-

1. CNN structure and processing
2. Kernel Size
3. Number of Epochs
4. Ideal Training Image Dimensions
5. Test Set and Validation Set Sizes
6. Ideal Weighing System (i.e. ADAM)

This CNN were tested for images tested from a close distance, medium distance and far distance and the accuracies were evaluated.