Learning to Find Features

Now that you've seen a number of feature extraction techniques, you should have a good understanding of how different objects and areas in an image can be identified by their unique shapes and colors.

Convolutional filters and ORB and HOG descriptors all rely on patterns of intensity to identify different shapes (like edges) and eventually whole objects (with feature vectors). You've even seen how k-means clustering can be used to group data without any labels.

Next, we'll see how to define and train a Convolutional Neural Network (CNN) that learns to extract important features from images.

Good work so far!