-1

Im trying to validate a PDF which is digitaly signed, and to extract the signers name. Is there any library for doing this job in python? So far I've just come across Itext but it is for Java. This is not a Computer Vision Problem as some of you are Sugesting, basically I want to replicate what Adobe Acrobat does to PDF, that it extracts certain information from the certificate. enter image description here

enter image description here

mkl
  • 90,588
  • 15
  • 125
  • 265
Brorx
  • 65
  • 7
  • signature presence detection with high rates of success using Custom Vision adding a simple bounding box to improve accuracy also works and can be done with a few lines of code.The only success I had with signature classification was when the customer was using Autopen (signature is consistent across every document). – Ram Jan 30 '20 at 12:05
  • @Ram but in my case it is encoded, does it matter? Ive seen this tool pdfsig in Poopler that seems to work, but it is C – Brorx Jan 30 '20 at 12:19

1 Answers1

-1

“signature presence detection”, which is do-able and can even be done using Custom Vision service. You can take that even further where certain forms have the expected printed name next to the signature. If the problem can be framed as a “detection” problem, this is more do-able currently. I have implemented this at a couple partners using simple techniques (bounding boxes of forms, Custom Vision NNs).

For a multiple reasons this is hard to do...As humans having multiple (versions) of their signatures etc. You can make a work-able demo, but from my experience it’s not going to be want to that really wants.

Ram
  • 2,459
  • 1
  • 7
  • 14
  • My problem is that the signature is not in the document as a visual thing, but it has information hashed or something into the PDF that can only be seen via Adobe Acrobat Reader and such softwares that tell you if the certificate is valid,(I will upload a picture in the main question so you can see) , so I have to extract this info and store the name of the company signing the PDF for proccesing – Brorx Jan 30 '20 at 12:54