Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Mistral OCR

mistral.ai

Post reply on HN
Print view View original

191–200 of 450 posts

Re: Mistral OCR

#191

Post by dml2135 »

kbyatnal wrote: post #140

We're approaching the point where OCR becomes "solved" — very exciting! Any legacy vendors providing pure OCR are going to get steamrolled by these VLMs. However IMO, there's still a large gap for businesses in going from raw OCR outputs —> document processing deployed in prod for mission-critical use cases. LLMs and VLMs aren't magic, and anyone who goes in expecting 100% automation is in for a surprise. You still n…

One problem I’ve encountered at my small startup in evaluating OCR technologies is precisely convincing stakeholders that the “human-in-the-loop” part is both unavoidable, and ultimately beneficial.

PMs want to hear that an OCR solution will be fully automated out-of-the-box. My gut says that anything offering that is snake-oil, and I try to convey that the OCR solution they want is possible, but if you are unwilling to pay the tuning cost, it’s going to flop out of the gate. At that point they lose interest and move on to other priorities.

Re: Mistral OCR

#192

Post by kashnote »

kiratp wrote: post #166

It's shocking how much our industry fails to see past its own nose. Not a single example on that page is a Purchase Order, Invoice etc. Not a single example shown is relevant to industry at scale.

Fwiw, they have an example of a parking receipt in a cookbook: https://colab.research.google.com/github/mistralai/cookbook/...

Re: Mistral OCR

#193

Post by hyuuu »

serjester wrote: post #114

This is cool! With that said for anyone looking to use this in RAG, the downside to specialized models instead of general VLMs is you can't easily tune it to your use specific case. So for example, we use Gemini to add very specific alt text to images in the extracted Markdown. It's also 2 - 3X the cost of Gemini Flash - hopefully the increased performance is significant. Regardless excited to see more and more compe…

gemini flash is notorious for hallucinating the output of the OCR, be careful with it. For straight forward, semi-structured, low page count (under 5) it should perform well, but the more the context window is stretched the more the output gets more unreliable

Re: Mistral OCR

#194

Post by sixhobbits »

Nice demos but I wonder how well it does on longer files. I've been experimenting with passing some fairly neat PDFs to various LLMs for data extraction. They're created from Excel exports and some of the data is cut off or badly laid out, but it's all digitally extractable.

The challenge isn't so much the OCR part, but just the length. After one page the LLMs get "lazy" and just skip bits or stop entirely.

And page by page isn't trivial as header rows are repeated or missing etc.

So far my experience has definitely been that the last 2% of the content still takes the most time to accurately extract for large messy documents, and LLMs still don't seem to have a one-shot solve for that. Maybe this is it?

Re: Mistral OCR

#198

Post by odiroot »

beklein wrote: post #174
odiroot wrote: post #94

Earlier quoted context omitted.

May I ask as a layperson, how would you about using this to OCR multiple hundreds of pages? I tried the chat but it pretty much stops after the 2nd page.

You can check the example code on the Mistral documentation, you would _only_ have to change the value of the variable `document_url` to the URL of your uploaded PDF... and you need to change the `MISTRAL_API_KEY` to the value of your specific key that you can get from the Le Platforme webpage. https://docs.mistral.ai/capabilities/document/#ocr-with-pdf

Thanks!

Re: Mistral OCR

#199

Post by hyuuu »

hubraumhugo wrote: post #123

It will be interesting to see how all the companies in the document processing space adapt as OCR becomes a commodity. The best products will be defined by everything "non-AI", like UX, performance and reliability at scale, and human-in-the loop feedback for domain experts.

I 100% agree with this, I think you can even extend this to any AI, in the end, IMO, as the llm is more commoditized, the surface of which the value is delivered will matter more

Re: Mistral OCR

#200

Post by adrianh »

s4i wrote: post #172

I wonder how good it would be to convert sheet music to MusicXML. All the current tools more or less suck with this task, or maybe I’m just ignorant and don’t know what lego bricks to put together.

Try our machine-learning powered sheet music scanning engine at Soundslice:

https://www.soundslice.com/sheet-music-scanner/

Definitely doesn't suck.

Post reply on HN