Google Research is showing a new way to use AI to read handwriting that could radically change the way machines convert what you put on paper into digital letters. The InkSight system transforms photographs of handwritten words into digital text by leveraging artificial intelligence without the need for any device as an intermediary.
The idea is to replace sometimes fallible optical character recognition (OCR) with AI that can emulate how humans actually learn to read, specifically by rewriting existing text to learn what full words look and mean. To do this, the researchers had to teach the AI to recognize and imitate human handwriting.
“Digital note taking is gaining popularity and offers a durable, editable and easily indexable way to store notes in vectorized form, known as digital ink. However, there remains a substantial gap between this form of note taking and traditional pencil note taking. and pencil. -Taking notes on paper, a practice that continues to be favored by a large majority,” the researchers explain in their article. “Our approach combines prior knowledge of reading and writing, allowing a model to be trained in the absence of large numbers of paired samples, which are difficult to obtain. To our knowledge, this is the first work that effectively represents handwritten text in arbitrary photographs with diverse visual characteristics and backgrounds.”
InkSight is more than just an alternative technique. Provides more accurate results in less than ideal circumstances. For example, if the photo was taken in low light, has partially obscured text, or has a confusing background when examined with OCR. The researchers found that humans could read 87% of text strokes made with InkSight. Two-thirds were good enough that people couldn't distinguish them from real writing; You can see below what it looks like when InkSight works.
Written by IA
If you like writing things by hand, InkSight has some potential benefits. Imagine writing by hand in a paper notebook and then showing the notes to your camera so they can be instantly searched and organized in context with previous notes on physical pages. If you're like me and have particularly messy handwriting, InkSight could help you turn your scribbles into typed text that's still exactly what you scribble.
On a larger scale, this could be a crucial tool for deciphering and converting handwritten texts over the centuries into digital format. Even when the text is in a language without much of a digital presence, InkSight could help preserve handwriting to help create training sources for those languages.
Google is not the only place where artificial intelligence tools are being developed to decipher handwriting. For example, Amazon's new Kindle Scribe improves the e-reader's ability to transform handwritten notes into readable text. There's also Goodnotes, a digital note-taking app that can read handwriting, and recently introduced handwriting editing tools that use its Goodnotes Smart Ink technology to convert handwriting into typed text. Added tools let you edit handwritten notes as if they were typed, including aligning notes, copying and pasting, and rearranging text to make it more logical.