Facebook is working on a new Artificial Intelligence model through which users can rewrite any text the way it is written in any picture. All that will be needed to copy the writing style is just one word. The first self-supervised AI model can replace text in existing images of both scenes and handwriting — in one shot. According to Facebook, this feature is expected to help with text translation, replication and picture editing.
The AI research project, titled TextStyleBrush, will also be submitted to a peer-reviewed journal.
- Facebook to tweak its 'News Feed' algorithm - Here's how it'll work
- Zuckerberg unveils Facebook's plans for new 'social audio' tools
- Facebook under pressure to scrap its plan for 'Instagram for Kids'
How does TextStyleBrush work?
According to a blog post from Facebook, the new tool works similar to the way style brush tools work in word processors, but for text aesthetics in images. "It surpasses state-of-the-art accuracy in both automated tests and user studies for any type of text. Unlike previous approaches, which define specific parameters such as typeface or target style supervision, we take a more holistic training approach and disentangle the content of a text image from all aspects of its appearance of the entire word box. The representation of the overall appearance can then be applied as one-shot-transfer without retraining on the novel source style samples," it added.
Still in research stage, when it becomes actual tool, it can power a variety of useful applications like translating text in images to different languages, creating personalized messaging and captions, and "maybe one day facilitating real-world translation of street signs using AR."
But what are its pitfalls?
But some fears of misuse are also there. For instance, quite evidently it can be used in picture editing to change the actual text. Of course, that carries enormous mischief value. Just imagine, a pic of a Prime Minister or President of a country from his/her college days carrying a placard. What if the words on the placard is changed and twisted to seem problematic? It can surely trigger a big outrage. And this tool can also be used to remove watermarks in images. That will be commercially critical.
Facebook has released its research paper on the project, and perhaps understanding the potential problems inherent in it, it says: "By publishing the capabilities, methods, and results of this research, we hope to spur dialogue and research into detecting potential misuse of this type of technology, such as deepfake text attacks — a critical, emerging challenge in the AI field."
