From the course: Deep Learning: Image Recognition

Unlock the full course today

Join today to access over 24,700 courses taught by industry experts.

Preprocessing and feeding data into your network

Preprocessing and feeding data into your network - Python Tutorial

From the course: Deep Learning: Image Recognition

Start my 1-month free trial Buy for my team

Preprocessing and feeding data into your network

“

- [Instructor] Next we will move on to some details with pre-processing and feeding data into your networks. So let's go ahead and open up 03_02_begin python file. So again, we will start with importing necessary libraries here as usual. Then we will normalize the data. We will then go ahead and visualize some of the images like we normally do. So after loading the dataset, we will go ahead, and print the shapes of the dataset to verify the loading. So this is a great time to stop and talk about data augmentation, and what is it, and why do we want to incorporate it in our code. So it involves creating new training examples by applying random transformations to the existing images. Well, why do we want random transformations to our images? Well, this helps the model generalize better by seeing more varied examples during the training. So the more different the data is, the much better models we can build with it. So we will go ahead and visualize the data before doing any augmentation…

Contents