These are MATLAB codes for traffic sign classification. Database used is German Traffic Sign Database, available for download here: http://benchmark.ini.rub.de/?section=gtsrb&subsection=dataset
Images from this database are very life-like and are challenging to classify. They are of different sizes and differently illuminated etc.
-
Linear Discriminant Analysis
-
Fisher's Linear Discriminant/Fisherfaces
-
Random Forests (in python)
algorithms to classify:
Raw intensity values from Images
Histogram of Oriented Gradients descriptors
To get these codes working, point to the correct directory containing dataset in readHOG.m, readTestHOG.m, readImages.m, readTestImages.m files
A detailed report containing results is availabe in report folder. References are also available in it.
#Note on Random Forests Implementation: We tried to implement our own randomforests class in python. As of the moment, it's not working and has to be debugged. Instead use RF_builtin.py to classify using scikit-learn's randomforest classifier class. As before, edit the file to point to the folder containing dataset.