Blog: Where are we surrounded by AR and AI?
ARchy everyone! Many people use smartphones, but how many of them
know how apps work, what technology is behind Instagram masks, how your
device determines the position of your head and the facial features? Let’s find
out how it happens with the help of AI and AR.
What is AI?
Many of us came across a “smart” news feed that allows you to select several categories that appeal to you (hi-tech, development, music) and, as a result, get a hotchpotch of news. Very often it is not an excellent one, but not a bad one either, just like a beginner cook would make. When will a beginner cook start making a great hotchpotch? How will he know that his hotchpotch is good? Only after he gets the necessary experience, feedback from people who tried his hotchpotch, and makes the hotchpotch multiple times.
AI (Artificial Intelligence), an emulation of human processes and consciousness, works in a similar manner. What does consciousness have to do with it, one may ask, but it plays an essential role in AI. A machine cannot have the same consciousness as a human; it is limited. Consciousness appears in neural networks. Imagine a two-year-old child, and you have to teach this child how to switch cartoon channels on TV. It is not an easy task.
You have to:
1. Show him what the remote looks like
2. Make sure that the child does not confuse the remote with anything else
3. Show which buttons to press to switch channels
Eventually, the child will master this process, and after a certain time, it will be usual practice for him. Neural networks work in the same way. The most common way to train a neural network is backpropagation. A neural network has input data and a result. Every action that leads to the final result has a specific value. In the case of the child, if any of the actions are devalued, he will not be able to switch channels. Thus, each action can be adjusted according to the desired result, which allows you to make the backpropagation algorithm based on the outcome. Don’t forget that showing the child different remotes, in other words training the neural network on different kinds of data, is not recommended because it can lead to “overtraining” — when a child sees a different remote, he will not classify it as a remote and will not be able to work with it.
What is AR?
Everyone who more or less follows the technological news in the world
heard about Pokémon GO
If you watched the movie “Terminator,” you would recognize this scene,
if not, I recommend to see it! Here you can see the world through the eyes of one of the main characters in the film — a machine from the future. Don’t worry! We are not there yet.
AR (Augmented Reality) is the visual augmentation of the world around us.
In this scene from “Terminator,” you can see how the real world is augmented, how the machine analyzes and displays information on the screen without the need to use a phone, laptop or glasses; and this trend is taking over the world. Unfortunately, technology has not yet evolved to the level we see in the film when characters can see the world like that. At the moment, the most popular way to use augmented reality is via smartphones that reveal the potential of AR.
Write in the comments below how many pokémons you managed to catch!
AR areas of application
AR is used in many fields:
5. Military technologies
We looked into how AR can be used for entertainment purposes, but it is unclear how it can be applied in other fields.
Leisure offers exciting things, for example, interactive and talking (just think about it!) wine labels
interactive products in magazines, which you can view from different angles, graffiti on the walls
the possibility to test fit the furniture for your room without leaving home
scanning marks on the map and much more. AR is everywhere!
In the field of education, AR is only gaining momentum and will eventually become very useful. At the moment, the functionality is limited to various hints with additional information from the teacher, which a student can see in his smartphone when pointing the camera at the textbook. In the future we expect smart classes, virtual assistants and much more to be promoted.
Now let’s get down to some serious topics, namely medicine and military technologies.
In the field of medicine mannequins and special programs for students are in high demand. For each mannequin, you can build a 3D model, which will display hints, show complex blood circulatory systems or the location of internal organs.
With this program, the student could carry out medical procedures with a mannequin under the control of the supervisor. But even during real surgery, doctors can rely on AR to see the possible risks.
In the area of military technologies targeting systems in various aircraft, tanks and drones have been used for quite a while. The military does not stop there and is planning to adopt helmets that will display buildings’ models, map information and the location of enemies. Such technologies will significantly simplify the life of military officers and also reduce costs.
All of the above helps to implement AI and display all sorts of information on the screen of a smartphone or any other device (e.g. AR-glasses).
Let’s look at the most common case in our everyday life — the masks in Instagram, Snapchat or any other application. We are all familiar with the situation when the app asks us to shake our head, stick out the tongue so that it can apply a cool mask or a mask effect. Computer vision makes this happen. It is a technology used by a machine to detect, classify and track an object.
A separate area in computer vision is image processing.
Image processing is the conversion of images done by performing mathematical operations on each pixel in the image.
The first thing that the machine does is recognizing the face of the person in the incoming image or video. At this stage, HOG (Histogram of Oriented Gradients) and SVM (Support Vector Machine) combinations are used. This is one of the key methods of finding face landmarks.
After that, the machine has to determine the facial features of the person: lips, eyebrows, nose, eyes, etc.
Finally, after all these actions comes image processing and applies the necessary mask to your face.
I like the Puppy mask, and you?
Now you understand what AI and AR are. These technologies surround us everywhere. AI is used much wider than AR. This is a big advantage for development in AR because AI has already accumulated an extensive database.
How do you like talking bottles? The fact that medicine, military, and education find the connection between these technologies very useful, proves its future benefits. Users catch pokémons around the corner of their house, take pictures of themselves wearing various masks, test fit furniture for their apartments without leaving home and much more. It seems that the life of the future has got much closer.