How can we read images/tables in PDF and csv/excel files using langchain? | Learn AI Together | Page 1

slim temple Dec 11, 2023, 7:03 AM

#

Is there a way to read images/tables that are in PDF, CSV, EXCEL files so that we can implement RAG and ask questions about it?

torpid orbit Dec 11, 2023, 10:41 PM

#

Yes you can use PyPDFLoader and it is well documented https://python.langchain.com/docs/modules/data_connection/document_loaders/pdf

PDF | 🦜️🔗 Langchain

Portable Document Format (PDF), standardized as ISO 32000, is a file format developed by Adobe in 1992 to present documents, including text formatting and images, in a manner independent of application software, hardware, and operating systems.

#

Load with PyPDF
Split document using RecursiveTextSplitter
Store them in a vector store (FAISS, Chroma, etc) with embeddings
Retrieve from the vector store

slim temple Dec 12, 2023, 6:37 AM

#

torpid orbit 1. Load with PyPDF 2. Split document using RecursiveTextSplitter 3. Store them i...

I did that but it doesnt work for complex pdfs that include tables and images

torpid orbit Dec 12, 2023, 8:47 AM

#

you can convert the pdf into markdown then use markdown document loaders, i dont know about the tables part, but it should work for the images

#

yes i can confirm it work with tables

#

https://github.com/facebookresearch/nougat

GitHub

GitHub - facebookresearch/nougat: Implementation of Nougat Neural O...

Implementation of Nougat Neural Optical Understanding for Academic Documents - GitHub - facebookresearch/nougat: Implementation of Nougat Neural Optical Understanding for Academic Documents

#How can we read images/tables in PDF and csv/excel files using langchain?