After giants like Google, the social media giant Facebook also steps in the sphere of AI based products and features. Not just that, the SM King also enters the room by setting a benchmark.
“TextStyleBrush is an AI research project that can duplicate the style of text in a photo with just a single word. You may modify and replace text in photographs with this AI model.
TextStyleBrush is the first self-supervised AI model that replaces text in photographs of both handwriting and scenery — in one shot — using a single example word, unlike most AI systems that can accomplish this for well-defined, specialized jobs.
Although this is a research project, it has the potential to open up new avenues for creative self-expression in the future, such as personalized messaging and subtitles, and it lays the framework for future advancements such as photo-realistic language translation in augmented reality (AR).
We hope that by releasing the capabilities, methods, and findings of this study, we may encourage discussion and research into detecting potential misuse of this type of technology, such as deep fake text attacks – a major, developing challenge in AI.”, as said by Facebook about the gamechanger!
Artificial intelligence-generated visuals have been growing at a rapid pace, capable of synthetically rebuilding historical events or altering a photograph to look like Van Gogh or Renoir. Now, using only a single word sample as input, we’ve created a system that can substitute text in sceneries and handwriting.
While most AI systems can accomplish this for well-defined, specialised jobs, creating an AI system that can grasp the intricacies of text in real-world scenarios as well as handwriting is a considerably more difficult AI issue.
It entails comprehending an infinite number of text styles for not just varied typography and calligraphy, but also for various transformations, such as rotations, curved letters, and deformations that occur when handwriting; background clutter; and image noise.
Facebook’s TextStyleBrush works in the same manner as style brush tools in word processors do, but for image text aesthetics. In both automatic tests and user studies, it outperforms state-of-the-art accuracy for all types of text.
Unlike prior systems, which defined certain factors such as typeface or target style supervision, we employ a more holistic training approach and separate the content of a text image from all aspects of the word box’s look.
On the innovative source style samples, the representation of the overall appearance can then be applied as a one-shot transfer without retraining.
Thus, the world awaits further updates on the Facebook’s AI-Imaging Gamechanger.