Black technology from Facebook makes virtual world easy

Facebook has never been idle since it entered the VR space. At the GPU Technology Conference, Facebook demonstrated its black technology neural net, which generates unique images based on textual descriptions.

We have all played the game of guessing idioms. This game, which has been popular in domestic and international social circles, has brought people closer to you. You just want to use your computer to draw the imaginary pattern in your head and you will see the network opposite. The domestic and foreign friends played a variety of answers. This simple and fun social game is very immersive.

But what Facebook shows us today is a magical neural network. It is no longer your picture, but the neural network automatically draws a pattern according to the text you enter.

What is even more interesting is that the neural network has an artificial intelligence system that will respond to your textual needs. For example, if you enter: beauty, neural network will immediately appear a picture of a beautiful woman, and then enter the beauty - beach, it will immediately appear on the beach of Bikini beauty. Neural networks have become the darling of artificial intelligence development. In short, it is an algorithm that tries to imitate the way the human brain operates.

Facebook's approach is to train tens of thousands of different pictures and train neural networks to connect specific words with specific pictures. It is like our human brain. When we enter text, we immediately think of the picture that the text will present. general. However, after the training of the neural network, the result of obtaining the answer will be very short, and even one day it may exceed the reaction speed of the human brain.

The combination of neural network and virtual reality

This technology is very important for VR. If the neural network can describe 3D objects according to words, it will speed up the creation of a VR content world!

Before the neural network, Google's DeepMind developed from a 2D game neural network into a 3D game neural network in just one year! Its famous product is to beat South Korea Go Master AlphaGo 4:1!


Neural network to solve 3D objects need to solve the problem

There are many obstacles to overcome in order to create neural networks that can generate 3D objects. The first is to adjust the algorithm to allow the neural network to recognize the computation from 2D to 3D content; second, the lack of 3D content makes 3D training difficult to perform; however, with advances in computer vision technology and the popularity of 3D computing, this will be promoted. The realization of technology. The important thing is that the popularity of VR technology will bring us more 3D content.

Looking forward to it, when a neural network can generate 3D content, it will become easy to create a virtual world.


The GLAREME desktop mixer is a 200-300W power source, powered by 100% copper motor, efficient and powerful. Paired with a 3 or 4L rotating bowl, the GLAREME desktop mixer can easily mix everything from whole-wheat pancake batter to oatmeal cookies to fluffy egg whites.Whether you have a sweet tooth or a sweet tooth, this entry-level mixer will do the trick.

Stand Mixer

Electric Stand Mixer,Small Electric Hand Mixer,Best Electric Stand Mixer,Small Electric Food Mixers

Jiangmen Taifai Technology Co.,Ltd , https://www.glaremecn.com