I didn’t realize that ChatGPT can also “read” text on images, until I tried to extrapolate some data from a screenshot of a publication.
In the past I used OCR via scanner, but considering that a phone has a better camera resolution than a 10 years old scanner, I thought I could use ChatGPT for more text extrapolation, especially from old documents.
Is there any variant of LLama or similar, that can work offline to get as input an image and return a formatted text extracted from that image? Ideally if it can extract and diversify between paragraphs and formatting that would be awesome, but if it can just take the text out of the image as a regular OCR could do, it is already enough for me.
And yes, I can use OCR directly, but I usually spend more time fixing the errors that OCR software does, compared to actually translate and type that myself… Which is why I was hoping I can use AI
Trend Engine: AI-Powered News & Trends 