Traffic Sign Recognition

The goals / steps of this project are the following:

• Load the data set (see below for links to the project data set)
• Explore, summarize and visualize the data set
• Design, train and test a model architecture
• Use the model to make predictions on new images
• Analyze the softmax probabilities of the new images
• Summarize the results with a written report

Rubric Points

Here I will consider the rubric points individually and describe how I addressed each point in my implementation.

---

Project code

The Jupyter notebook

Data Set Summary & Exploration

1. Exporation of the dataset

• The size of training set is 34799
• The size of the validation set is 4410
• The size of test set is 12630
• The shape of a traffic sign image is (32, 32, 3)
• The number of unique classes/labels in the data set is 43

2. Include an exploratory visualization of the dataset.

Below are random examples of signs per each class and quantities of images per each sign plotted

Design and Test a Model Architecture

1. Preprocessing

In regards of preprocessing I've used OpenCV. Specifically I've used Normalize function to scale pixel's values from 0 to 1

2. Neural network architecture

My final model consisted of the following layers:

Layer	Description
Input	32x32x3 RGB image
Convolution 5x5	1x1 stride, VALID padding, outputs 28x28x6
RELU
Max pooling	2x2 stride, VALID padding outputs 14x14x6
Convolution 5x5	1x1 stride, VALID padding, outputs 10x10x16
RELU
Max pooling	2x2 stride, VALID padding outputs 5x5x16
Flatten	Input 5x5x16, output 400
Fully connected	Input 400, output 84
RELU
Fully connected	Input 400, output 43

3. Training and Accuracy

Model was trained with the following characteristics:

• Device: AWS GPU xLarge instance
• Optimizer: AdamOptimizer
• Learning rate: 0.001
• Epochs: 30
• Batch size: 128

Accuracy on validation set: 0.933

I was using LeNet model architecture due to it's proved efficiency in terms of images classification

Test a Model on New Images

1. Traffic signs downloaded from web

Here are five German traffic signs that I found on the web:

2. Prediction results

Here are the results of the prediction:

Image	Prediction
No entry	No entry
No passing	Vehicles over 3.5 metric tons prohibited
Speed limit (20km/h)	Speed limit (20km/h)
Keep right	Keep right
Pedestrians	Pedestrians

The model was able to correctly guess 4 of the 5 traffic signs, which gives an accuracy of 80%.

3. Predictions probabilities