cool hit counter Painter in danger Microsoft AI painting robot can "read text and paint"_Intefrankly

Painter in danger Microsoft AI painting robot can "read text and paint"


Microsoft has launched a new artificial intelligence painting robot, according to foreign media reports. The robot is able to create images based on textual descriptions of things and can also add details on its own that are not described in the text.

Microsoft launched a new artificial intelligence technology, a drawing bot, on January 18. The robot is able to create images based on textual descriptions of things, and it can add details to those images that aren't in the text, showing that artificial intelligence has an imagination of its own. Whether it's a plain idyllic scene or some absurd image, such as a floating double-decker bus, the robot is able to draw it, the researchers said.

If you search for 'bird' using the Bing web page, a picture of a bird will come up," according to Xiaodong He, a principal investigator at Microsoft's Deep Learning Technology Center at Redmond Regional Research Labs in Washington, D.C. But this drawing robot, will draw you a picture of a bird from scratch little by little. The bird probably doesn't exist in the real world, it's just a computer's imagination of the bird. "

Microsoft said the researchers were training the bot with many sets of data sets of images and captions so that it learned how to match text to images. For example, when the robot learns to draw a bird, it learns what a picture of a bird looks like when the title shows "bird". This is the fundamental reason why we believe that machines can learn," He said. "

What's particularly interesting is that thanks to all that training data, the drawing bots possess a bit of basic common sense and imagination, and fill in the blanks when no specific details are provided. Using the example of a bird, the robot will often draw a bird standing on a branch, even though the text doesn't mention this detail, and that's because that's what most of the images it was initially fed were like.

The technology of this drawing robot consists of a two-part mechanical learning model, one that generates images from textual descriptions and another that determines the authenticity of the generated images based on the textual descriptions. The two combine to create a higher quality image. According to a recent research paper, this new robot is nearly three times better in image quality compared to previous technology.


Recommended>>
1、Sanic Tutorial 1 Quick Start
2、What an eyeopener to show you the balancing shaft machining of a highend automotive shaft
3、Revealing the secrets behind Thai latex pillows Worth the price
4、Perceive Security to Wise the Future Xinhuasan Releases Latest Security Situational Awareness System
5、The Ministry of Education has spoken By 2020 the construction of 100 artificial intelligence X composite specialties

    已推荐到看一看 和朋友分享想法
    最多200字,当前共 发送

    已发送

    朋友将在看一看看到

    确定
    分享你的想法...
    取消

    分享想法到看一看

    确定
    最多200字,当前共

    发送中

    网络异常,请稍后重试

    微信扫一扫
    关注该公众号