search

piotrdelikat / fet-data-extractor

MIT License
0 stars 0 forks source link

Try to input PDFs instead of PNG #2

Open fl4p opened 1 week ago

fl4p commented 1 week ago

Try to send the PDF (or URL to download) to the LLM. Also flux.ai API might be worth to explore.

Note that some datasheets are images without text:

fl4p commented 1 week ago

Some infineon datasheets have text, when copied or extracted is a useless string.

Example: https://www.infineon.com/dgdl/IPB019N08N3_Rev2.1.pdf?folderId=db3a304313b8b5a60113cee8763b02d7&fileId=db3a30431add1d95011ae87fdf90569f

#   !  !  
"%&$!"#D  # : A 0<& <,9=4=>: <
6LHZ[XLY
Q#451<6? B89786B5AE5>3 ICG9D3 89>71>4CI>3 B53
Q( @D9=9J54D53 8>? <? 7I6? B=? D? B4B9F51@@<93 1D9? >C
Q H3 5<<5>D71D53 81B75H' 
9H"[Z#

claude.ai however, successfully extracts tabular data.

fl4p commented 1 week ago

Special Data Sheets: