Microsoft Company recently unveiled its newest AI – “a drawing bot
The Microsoft recently unveiled its newest AI – “a drawing bot”. Yes and this one is an artist! Researchers from Microsoft explained that this bot is capable of creating images from text descriptions of an object. In addition to that, it will also add details in the result images which weren’t part of the text description. Here’s how it works! You go to Bing and type ‘bird’ on the search bar and you get images of birds. Now, the twist is that these images which are displayed are created by the computer from scratch. It creates the entire image pixel by pixel from its imagination based on the text you type. The birds in these images, however, may not exist in the real world but are just an aspect of the computer’s imagination of birds. So, if you type “blackbird”, chances are that you would see the bot paint a black bird sitting on a branch and it might even add the blue sky for a clearer representation. These details weren’t included in your text but are in the image. This hints that AI has some kind of the imagination of its own.
Microsoft’s scientists likewise said that this bot was prepared on datasets of inscriptions and matched pictures
The team of researchers in Microsoft explored the options understanding a computer’s vision and natural language processing for over 5 years before introducing “Drawing Bot” to the world. Microsoft’s previous AI – the ‘Caption Bot’ which automatically writes captions for images was the first release of imaginative bots. When inquired, the researchers mentioned that the image generation is comparatively a tougher feat to achieve than captioning as the former requires intricate machine learning algorithms to imagine missing parts of the text and accordingly drawing a picture. Microsoft’s scientists likewise said that this bot was prepared on datasets of inscriptions and matched pictures. The innovation at the centre of this bot is known as Generative Adversarial Network (GAN) which comprises of two machine learning models cooperating to deliver the best outcomes. The preparation models assisted the Microsoft’s Drawing Bot to map words with its visual form.
Researchers also believe that text-to-image generation technology will be of use in practical applications
With this new launch, the previous state-of-the-art technique for text-to-image generation reaches a newer standard. Researchers also believe that text-to-image generation technology will be of use in practical applications such as an assistant to painters or interior designers. Despite the fact that this innovation is still being worked and that it has to undergo further changes and upgrades, things appear to be brilliant for Drawing Bot. You can literally control and enhance the technology as recent tests proved that the bot can also paint an image resembling absurd descriptions that you enter and that it learns all the way. As a test, the research team asked the bot to draw a “red double-decker bus floating on a lake” and it generated a blurry image that somehow represented a boat with two decks and a double-decker bus on a lake surrounded by mountains. With more powerful algorithms, researchers see a possibility of this technology to be used in generating animated films based on screenplays, voice-activated photo-refinement and so forth while taking off a fair share of the manual labor involved. Let us wait and see how things turn out for Drawing Bot.