DeepTraffic

An online repository for the school project DeepTraffic.

This repo contains various bits and pieces used during the project:

• Android directory contains an Android studio project for an app that is a part of this assignment. Read more about it in it's own README.
• Models directory contains models trained during the project. It also has it's own README.
• Utils directory contains pieces of code not written by the author but used for the project.
• The rest of the repository contains other utilities but most notably the label_image.py function.

How to use label_image.py

Run the script from the console to label individual images.

Example usage:

python3 label_image.py
# to label the example_sign.jpg

python3 label_image.py --image=path/to/image
# to label a different image

python3 label_image.py --graph=Models/traffic_simple.pb --input_layer=flatten_input --output_layer=dense_2/Softmax
# to use the simple model

python3 label_image.py --graph=Models/retrained_MobileNetV2.pb --input_layer=module_apply_default/hub_input/Mul --output_layer=final_result --input_height=96 --input_width=96 --grayscale=False
# to use the retrained MobileNetV2

python3 label_image.py --help
# for more information

Prerequisities

This project uses the following libraries:

• Tensorflow
• matplotlib

You need to install these in the same Python environment for it to work properly.