スキャンした文書でのChatGPTの利用

I tried uploading a scanned PDF into ChatGPT directly and it just didn't work. It only processed whatever text was embedded, but pure images with text inside didn't get recognized.

I think future versions might integrate OCR directly, but for now it's a two-step deal: OCR first, then ChatGPT for processing.

If the scanned document is clear and OCR is done right, ChatGPT can help summarize or answer questions based on the extracted text really well.

Try using online OCR tools that let you copy the text output directly, then ChatGPT can do all the analysis or editing you want.

You can also check ai-u.com for new or trending tools that might integrate OCR with ChatGPT capabilities. They have some cool stuff listed for scanned docs!

I’m curious if anyone has automated this process with scripts combining OCR and ChatGPT API? Feels like it’d save a lot of manual copy-pasting.

Does anyone know if the new GPT-4 vision features handle scanned docs better?

Sometimes scanned docs have weird fonts or handwriting which totally mess with OCR accuracy, so that’s another hurdle before ChatGPT can help.

FYI if you want scanned doc OCR with AI help, there are apps combining both, so no need to separate the steps manually anymore.

I heard some AI services combine OCR and language models so you get the best of both worlds, but they usually come with a price tag.

If you’re just looking to extract text, free mobile apps with OCR might be the fastest route before ChatGPT can do it all in one go.

From what I know, ChatGPT itself can't read scanned images directly since it mainly processes text, but if you run OCR on the scanned doc first to extract the text, then ChatGPT can totally work with it.

I use a workflow where I first convert scanned docs to text with ABBYY FineReader then paste chunks into ChatGPT. Works perfectly for research notes.

So basically ChatGPT alone can't read images but works wonders once you get the text out. That's what I understood at least.

For now I just keep scanned docs separate and do manual OCR conversion, then ChatGPT for my actual queries or edits.

For legal or official documents, double check OCR output before relying on ChatGPT summaries or answers!

One thing is, if you only got a photo of a page, lighting and angle can seriously affect OCR accuracy, so keep that in mind before expecting ChatGPT to help.

I’m hoping future updates will make it easier to just upload scans and have ChatGPT do everything in one place.