The new AI technology from Microsoft is programmed in such a way that it can create images from text descriptions. The technology pays attention to particular words from the text description and uses a technique of focused attention to creating images of high quality. The Focused attention technique has allowed about three times increase in image quality compared to previous text-to-image generation. Xiaodong He, the principal researcher in Deep Learning technology at Microsoft says that the images produced by the bot are created pixel by pixel, from scratch. Xiaodong He says “These birds may not exist in the real world — they are just an aspect of our computer’s imagination of birds”. Well, the bot which converts text into images us using a technology named GAN (Generative Adversarial Network). Generative Adversarial Network uses two machine learning models – one generates images from text and the other one uses text to judge the authenticity of generated images. Both of the machine learning models work together to produce perfect images. The researchers say that the AI artist bot can generate any kinds of images like floating double-decker bus, grazing livestock, etc. Moreover, each of the generated images had details that were absent from the text descriptions. For more detailed information, you need to read the research paper. So, what do you think about this? Share your views in the comment box below.
