Training a Model

Training a Neural Network

To train a computer vision neural network, we typically provide sets of labelled images, which we can compare to the predicted output label or recognition measurements. The neural network then monitors any errors it makes (by comparing the correct label to the output label) and corrects for them by modifying how it finds and prioritizes patterns and differences among the image data. Eventually, given enough labelled data, the model should be able to characterize any new, unlabeled, image data it sees!

A training flow is pictured below. This is a convolutional neural network that learns to recognize and distinguish between images of a smile and a smirk.

This is a very high-level view of training a neural network, and we'll be diving more into how this works later on in this course. For now, we are explaining this so that you'll be able to jump into coding a computer vision application soon!

Example of a convolutional neural network being trained to distinguish between images of a smile and a smirk.

Gradient descent is a a mathematical way to minimize error in a neural network. More information on this minimization method can be found here.

Convolutional neural networks are a specific type of neural network that are commonly used in computer vision applications. They learn to recognize patterns among a given set of images. If you want to learn more, refer to this resource, and we'll be learning more about these types of networks, and how they work step-by-step, at a different point in this course!