Extract Text from PDFs & Images for LLMs Using Python

This video aims to provide a few technics to efficiently extract text from any type of document.
After completing this tutorial, you will have a clear idea of which tool to use depending on your use case.

Source code:

Medium Article:

Connect:

Support:

About me:
Hey there! My name is Zoumana. I got my Bachelors's in Mathematics and Computer Science, and my Master's in Computer Science with concentrations in Machine Learning & Data Science in Paris. I hold a second Masters's in Data Science & Business from Texas Tech University. I want to share my expertise with you and help you grow 📈 in your career as Data Scientist so that you can become an expert yourself.
#python #datascience #ocr #langchain #machinelearning

[eloghosaikponmwoba]

Thank you for this video. I needed this information at this very moment.

[techwithzoum]

Please subscribe and like the video to help me keep motivated to make awesome videos like this one. :)

[AbdulAhad-Family]

Fantastic tutorial, so much simplified ...great job

[kenchang]

Excellent tutorial, thank you. Your video was referenced on AI Jason's channel and I am glad he did.

[SMCGPRA]

The voice is very clear and crisp for non-english viewers to understand. Content is very excellent and explained exquisitely. Could you let us know if tables in the pdf or word doc, using RAG and prompt can we able to join tables, filter data from tables and other operations in Gen AI?

[rashmikasaha]

Hey, im getting an error
in isfile(path)
28 """Test whether a path is a regular file"""
29 try:
---> 30 st = os.stat(path)
31 except (OSError, ValueError):
32 return False

TypeError: stat: path should be string, bytes, os.PathLike or integer, not JpegImageFile

can you help me with a fix?

[cbylnrh]

مااسم صفحة الانترنت التى تكتب فيها الاكواد

[svbkjfj]

Hi there, I plan on using the EasyOCR Library for some sensitive Documents, Is it safe, like can any data Leaks Occur, also Is there any Documentations of the Library I can refer to ?

Thanks !!

[shooby]

I run into the following error when I try langchain's UnstructuredImageLoader:

TypeError: stat: path should be string, bytes, os.PathLike or integer, not JpegImageFile

[piyushchhawachharia_]

I used a third party tool API to extract text and tables but the image part is not working for it. It’s not even recognising the images. If I just use the python libraries instead that will recognise the image and I can save them to other folders and later work on it but the extraction of tables won’t work I guess for python libraries.

[thibauteka]

Thanks so much. can we have an example for a data extration from a table on an image ?

[vivekpatel]

@zoumdatascience Can we give the image output with text based on the questions?

[Abhi_interiors]

which among these has the best accuracy?

[SuiGio]

Important question: If a pdf does have a picture in it, when converting to picture, firstly, is that picture added as text or is it skipped?
Secondly, is there a way to *know* that the extracted text is coming from an image within the pdf? Some sort of metadata at least to get that info?
Thanks for the video, nice content with overall breadth, wish you could answer my question.

[anubhav]

I am getting list index of range for langchain. Can you suggest sometime there

[susmitsekhar]

Great work. can we extract information from charts like histogram/barplot ?

[cbylnrh]

كل الاساليب لتحويل الملفات فشلت ولا اعرف السبب

[ibrahimkouma]

Bonjour êtes vous malien? super tutoriel merci pour le partage

[hajarabdullah]

that was very useful for me, but i faced one struggle in langchain unstructuredimageloader it say its not allowed PNGImage? i don't get it, I try to resolve it but i couldn't, pls if there is any way to contact you, I'll appreciate . thank you