So far in this book, we have only considered still images. However, in this chapter, we will introduce the techniques that are applied to video analysis. From self-driving cars to video streaming websites, computer vision techniques have been developed to enable sequences of images to be processed.
We will introduce a new type of neural network—recurrent neural networks (RNNs), which are designed specifically for sequential inputs such as video. As a practical application, we will combine them with convolutional neural networks (CNNs) to detect actions included in short video clips.
The following topics will be covered in this chapter:
- Introduction to RNNs
- Inner workings of long short-term memory networks
- Applications of computer vision models to videos