【谷歌“刀锋战士”图像AI可锐化任何照片】

【谷歌“刀锋战士”图像AI可锐化任何照片】

“刀锋战士”是谷歌的最新人工智能软件,它基于两种神经网络和机器学习,可挖掘隐藏于模糊图片中的细节,绘制出8x8像素化的图像,继而通过搜索筛选和原图各部分细节匹配度最高的、高分辨率的各种图像,自动弥补原图丢失的细节。这是AI对基于特征脸的人脸识别方法的成功实践。

http://www.dailymail.co.uk/sciencetech/article-4201838/Google-reveals-photo-enhancement-tool-sharpen-snaps.html

‘Zoom in… now enhance’: Google reveals 'Blade Runner' photo AI that can sharpen any image

  • System uses two neural networks and machine learning to sharpen pictures
  • Networks search for high-resolution images that it believes matches the source
  • One network maps the 8 x 8 pixelated image against other high-resolution
  • Other uses PixelCNN to add high-resolution details to the pixilated source
  • Final image is the combination of the two neural networks' pictures  

Google Brain’s latest software can create sharpen images from a pixelated source.

The system combines two neural networks and machine learning to guess what details lay hidden in the blurry picture.

Once the system is fed an 8 x 8 pixelated image, the networks search for high-resolution images that it believes matches the source's content - and adds the missing details.

The system combines two neural networks and machine learning to guess what details lay hidden in the blurry picture. Once the system is fed an 8 x 8 pixelated image, the networks search for high-resolution images that it believes matches the source's content

The system combines two neural networks and machine learning to guess what details lay hidden in the blurry picture. Once the system is fed an 8 x 8 pixelated image, the networks search for high-resolution images that it believes matches the source's content

HOW DOES IT WORK?

The team at Google Brain has developed a system that is capable of making out details of a pixelated source.

The system combines two neural networks: conditioning network and prior network.

Condition network is the first part of the process that maps the 8 x 8 pixelated image against other high-resolution that it believes contain similar features.

Prior network uses  PixelCNN to add high-resolution details to the pixilated source.

The two networks will then combine the images they have chosen, in order to create the final picture.

‘Zoom in… now enhance’ is a phrase used by many script writers to reveal a twist in a movie plot and the technology used in these movies.

It has been used in a range of movie including Blade Runner and Star Trek.

But now, Google has devised a method that brings the sci-fi technology to the real world.

Ars Technica, the first to report on the subject, notes that these neural networks are not capable of enhancing or producing the original image.

But, the final result is what the system thinks it may have looked like - it makes an 'educated guess'.

The conditioning network kicks off the process by mapping the 8 x 8 image against other high resolution images.

It then downsizes the high-resolution images to the same 8 x 8 pixel size in order to find ones that could match the features.

This image is a rough draft for the second neural network, which works its magic to add more details to the blurred picture.

Called prior network, this part uses PixelCNN to add high-resolution details to the8x8 source.

These neural networks are not capable of enhancing or producing the original image. But, the final result is what the system thinks it may have looked like - it makes an 'educated guess'. The three last columns were images produced using Google Brain's software


Comments are closed.



无觅相关文章插件