• AmbiguousProps@lemmy.today
    link
    fedilink
    English
    arrow-up
    16
    arrow-down
    1
    ·
    2 days ago

    You could’ve just looked for off the shelf OCR software and it would probably be better, no LLM needed. OCR has been around for far longer than the current LLM bubble.

        • jaykrown@lemmy.world
          link
          fedilink
          English
          arrow-up
          1
          arrow-down
          1
          ·
          17 hours ago

          Yea you could argue semantically that using an LLM to turn text in an image into machine readable format falls within “Optical Character Recognition”. I was referring specifically to OCR algorithms like Tesseract (pytesseract) and EasyOCR.

          • ganryuu@lemmy.ca
            link
            fedilink
            English
            arrow-up
            1
            ·
            5 hours ago

            you could argue semantically

            No. There’s nothing to argue there, it’s the definition of OCR.

            Also, do you believe that LLMs found a new, novel way of doing OCR? That’s not how they work, LLMs don’t invent, they don’t innovate, they’re simply unable to do that. What they do, when they work correctly, is that they use already known and established techniques and tools. So to quote your top comment in this chain:

            Skill issue