Computer vision is everywhere — from facial recognition, manufacturing, agriculture, to self-driving vehicles. Today you’ll enter the world of modern computer vision with a hands-on example. You will learn how to detect license plates with the YOLO algorithm.

Why do you need YOLO? What’s wrong with traditional computer vision approaches? Well, my friend, I have three words for you:

They all suck.

Most articles use old-school computer vision methods to detect the plate. Thresholding and contour detection isn’t enough in late 2020. Their algorithms work on images within the article due to hardcoded values, but fail to generalize. The result is an unusable garbage model.

You are not here to build garbage. After reading this article, you will have a robust model for detecting license plates. It will work every time and in every weather condition.

The article is structured as follows:

• Data gathering and preparation
• Model training
• Model evaluation
• Conclusion

Before we get started, I would like to take a moment to give a shoutout to all YOLO object detection tutorials, especially this one. Some of the solutions you’ll see throughout the articles are based on them.

#machine-learning #python #data-science #opencv #yolo