Facing problem in extracting aadhar card using Document understanding

Help Studio
1

Hi all,

I have doing a self project on extracting aadhar card details.I am using document understanding.Whether i use keyword extractor or machine learning its unable to extract name of the person.

The images are clicked from phone and i am using UiPath document ocr

In some images it is like
First name: Name

Some its like
Name

Thanks!

2

@Jai_Pande

Have you tried with pdf activities

check it once its better if the fields are not constant

or you need to use different extractor for different Aadhar card types

cheers

3

Actually i want to make a project with DU only, so can you kindly help me out

4

Also i need to ask do i have to use separate DU activites as for ex there i a single aadhar card but there are two images of it front and back…front contains name etc…and back contains address

5

@supermanPunch need help man!

6

Hi @Jai_Pande ,

Could you let us know upto what stage have you performed the Process Design ?

What is the Model used ? If DU Model, what was the number of documents used for Training (Labelling+Pipeline Run) ?

7

There are 5 docs available. 1 Aadhar card and 1 pan card. Each has 2 documents namely:- Front and Back

8

@Jai_Pande

Then you need to use two extractor one for Pancard other for Aadhar Card

cheers

9

yes but aadhar card too has two pages front and back…and when i am extracting the front page there is data of back page also in present validation station

10

@Jai_Pande ,

Not yet fully able to understand the implementation done. Could you let us know the progress of the implementation upto now as previously mentioned ?

11

Oh! the model used is Document Understanding with Flowchart brother, actually i am unable to extract the name its showing something else in validation station

12

@Jai_Pande

try with form extractor or Regex based extractor

cheers

13

@Jai_Pande

can you confirm if you have performed labeling the data and trained the model.

Cheers

14

guys its always showing Yes in field of name, that the major problem here

15

i am using trainer after present validation station , the workflow is as follows:- Taxo…digitization…classify…extraction…present validation station…and then classifier trainer

16

using form extractor

17

@Jai_Pande

try once with regex based extractor as you are not getting the appropriate values using Form extractor

cheers

18

ok and should i create separate document types in taxo for aadhar card front and back??

19

@Jai_Pande

see basically we use one taxinomy for one type of pdf,as aadharfront and back as combined will called as a single pdf

for pan card you need to add another one in the taxinomy

20

okay i have two separate images of aadhar card brother…front and back