So, there we saw that the classifier that we built for classifying fashion, like using convolutions were able to make it more efficient and to make it more accurate. I thought that was really, really cool, but it's still very limited in the scenario because all of our images are 28 by 28 and the subject is actually centered. And while it's a fabulous dataset for learning, it's like when we start getting into real-world images and complex images that maybe we need to go a little bit further. What do you think? I think it's really cool that taking the core idea of a confinet allows you to implement an algorithm to confine not just handbags right in the middle of the image but anywhere in the image, so it could be carried by someone on the left or the right of a much bigger and, say, a one-megapixel image. This is 1000 by 1000 pixels. Also for many applications rather than using grayscale, want to use color images- All right. And the same core ideas but with a bigger dataset, bigger images in similar labels lets you do that. All right. So, the technique that you're learning in this, is really really helping you to be able to succeed in these more real-world scenarios. So, I know you've been working on a dataset on horses- Yeah. And humans. Yeah, that's been a lot of fun. I've been working on a dataset that's a number of images of horses and they're moving around the image and they're in different poses, and humans in the same way and diverse humans male, female, different races, that kind of thing to see if we can build a binary classifier between the two of them. But what was really interesting about this is that they're all computer-generated images, but we can use them to classify real photos. I had a lot of fun with that. So, I think there'll be a fun exercise for you to work on as well. And if you're ever wondering of these algorithms you're learning whether this is the real stuff, the algorithims you're learning is really the real stuff that is used today in many commercial applications. For example, if you look at the way a real self-driving car today uses cameras to detect other vehicles or pedestrians to try to avoid them, they use convolutional neural networks for that part of the task, very similar to what you are learning. And in fact, in other contexts, I've heard you speak about using a convolutional neural network. To take a picture, for example. Yeah, we can take a picture of a crop and try to tell if it has a disease coming. So, that was really cool. Oh, thank you, thank you. That's really fun. So, in the next video, you'll learn how to apply convolutional neural networks to these much bigger and more complex images. Please go on to the next video.
