Road Signs Recognition using multilayer perceptron neural network algorithm
Following software was tested and executed using Anaconda 5.3.0 on Windows.
get_model.py expects image files with road signs under dataset directory
During develompent phase we are using Belgian Traffic Sign Dataset linked in references, specifically:
- BelgiumTSC_Training (171.3MBytes)
- BelgiumTSC_Testing (76.5MBytes)
Example format after unpacking:
datasets/BelgiumTS/Training/
datasets/BelgiumTS/Testing/
- Install Anaconda
- Configure your software to use Anaconda enviornment
- Unpack dataset to project directory as explained earlier
- Cd to project directory
To train the classifier:
python get_model.py
To classify a single image, using a pre-trained classifier:
python classify.py
To start a graphical interface for classifying images:
python classify_gui.py
To run road signs classifying accuracy tests with added noise or blur (parameters configured inside .py file):
python distort_and_classify.py
Belgian Traffic Sign Dataset - BelgiumTS
Reference Belgian Traffic Signs - OpenStreetMap Wiki