In this tutorial we will be using binary Logistic Regression for training a red color classifier in Python.

In order to keep the tutorial relatively short, I have divided it into two parts — Part I: Creating a color classifier using logistic regression. Part II: Detecting stop signs using the binary mask obtained from the color classifier.

The first step of any machine learning problem involves having access to a good training dataset — in our case, we require images that consist solely of the color red (positive samples), and images containing colors other than red (negative samples). I was able to create a dataset which had a diverse pool of red images extracted from the stop signs themselves under various lighting conditions. These lighting variations will allow us to train a relatively robust color classifier. Colors other than red were included in the training set as negative samples. You can find the dataset here on my GitHub: https://github.com/MariaHarris24/StopSign_detection.

Before I illustrate the coding part of this tutorial, we first need to understand the underlying principles of Logistic Regression and the math involved in training it.

#artificial-intelligence #machine-learning #classification-algorithms