Unanswered

Pulled data from multiple pdfs into one excel file

(0) Share

Report

Posted on by AbdulWasay1

Hi Everyone,

I am trying to pull data from multiple pdfs into excel. I created my AI Hub and trained the model and got 96% accuracy.

After I created my flow and tested it, the data was coming in accurately. Upon wider scale testing with 50 pdfs, total 912 records were extracted out of which 40 records had the column amount blank though it was populated in PDF.

There were other records with the same description but they were accurately extracted from the PDF to excel.

Can anyone please advise how can I increase the accuracy so that no value goes unextracted?

Categories:

AI Builder

I have the same question (0)

All responses (2)

Answers (0)

NathanAlvares24 1,714 Moderator on at

Like (0)

Report

Hi @AbdulWasay1 !

Actually 96% accuracy is pretty good. Very low chance of blank data.

Have you tried doing a 'Quick Test' on those PDFs it extracted that blank data? Better to check first. And if it does not extract, try adding that PDF as part of the training data and re-train your model.

Sometimes I also encounter some PDFs or Images which cannot extract the text in AI Builder maybe because of low DPI.

Do let me know the updates once done.

Regards,
Nathan Alvares

Was this reply helpful? Yes No
AbdulWasay1 4 on at

Like (0)

Report

Hi NathanAlvares24

I did the Quick Test on those PDFs and it extracted the blank data.

After that, I added the PDFs as part of the training data (added 5 more pdfs) and trained the model and my accuracy dropped for 96% to 46%.

I added 5 random PDFs to it and it dropped then from 46% to 43%.

I am not sure what just happened here.

Regrads,
Abdul Wasay

Was this reply helpful? Yes No