Hello everyone. In this video, we're going to talk about two image to Image Search products that are currently in production. One of them is Google Images. And the second one is theoria.ai, which is smaller company, but very, very impactful in this field. So if you use Google Images before, you probably noticed this small picture right here. And I'm not sure if you use it before, but there is an option if you click it here, you're going to end up with this pop up where you can upload an image and it will try to search similar images to the one that you uploaded.

So if you go to the image to the browser right here, we are going to select an image. Just to know though I'm currently in Serbia, so everything here as you can see is in Serbia language. So if I go to browse, we'll end up with this test folder well test folder from our data set that we are going to use In our course, and I'm going to use these images so we can compare our end to end pipeline that we are going to implement in this course and what currently Google uses. So if you end up, for example, upload this truck, you're obviously expecting to get similar trucks, right? So let's see what we are going to get. Okay, as you can see, it recognizes buildings and yeah, there is some blue color in an image, some brown I would say.

And, indeed, Google recognize those colors. And as you can see, if you go to visual close images, you will see that better on colors, but none of these images are actually trucks. So Google didn't achieve what we try to achieve. Let's try with another image. Let's go With the image, for example of a dog. And yeah, indeed, this, this is a great example where Google image search works perfectly.

So as you can see, we did recognize old German Shepherd, which we are not sure if this is really German Shepherd, because very It is very small image. But yeah, it is a dog, and we try to get a dog. And as you can see, if you go to visual similar images, there are a lot of examples that are indeed dogs. And basically, all of these colors are what we had in our original uploaded image. So in this example, Google did a great job. Let's try with another example.

This time, let's go with bird Okay, in this example, you can see that Google works perfectly. It did recognize bird, it tried to recognize even further with a specific breed of a bird, which we are not going to do in our course we are going to work with just birds and high level classes in general. So you can see that Google is doing a great job and yes, it is going to make mistakes because it is pretty generalized one, it should work with any image that you upload. As I promised, let's go to our second example, which is theoria.ai. Styria is a major company in Croatia and they are doing several things and one of them is image to Image Search. And to be honest, the first time that I heard about image image search are from them.

So they are working on several applications for Imaging Research. One of them is for the View hub and fashion cam, which is one very big company from Austria. And they built image search in their application which is similar to eBay. And when you take a shot for example of the shirt, it is going to try to recognize and find similar searchers in the store. And basically their product is very profitable and received several Nvidia rewards. And as you can see, you can build a great product to just from image to Image Search.

It is specialized it is just for a company. So it is not generalized as a Google one, but it works perfectly and very fast. Actually, their image to serve pipeline is the fastest in the world right now. That would be it for this video. If you have any questions or comments so far, please post them in the comment section. Otherwise, and see you in the next tutorial.