Engine2 missing simple numbers on ID Card OCR


Hello Dear OCR Space, I am developing an app where im trying to extract some fields from for example an ID Card. I am trying to extract three fields (name, lastname and date of birth). The API mostly works well, but on some cards it cant read simple digits, like “1.”, or “2.”, and this is how these
names are extracted. A nordic id-card looks like “1. Surname”, “2. name”, “3. date of birth”. Do you have any tips to make this better? I have tried
the different engines and scale/orientation etc. From this example card it doesn’t retrieve the “2.”, just the letters after it.
Since the upload limit is 1mb, maybe it would work better with the pro tier and larger images?
Much appreciated,
Kuken666

This is the parsed text:
ParsedResults: [
{
TextOverlay: [Object],
TextOrientation: ‘0’,
FileParseExitCode: 1,
ParsedText: ‘N\n’ +
‘FORERKORT\n’ +
‘NORGE\n’ +
‘1. NORDMANN\n’ +
‘KARI\n’ +
‘3. 1959-07-02-\n’ +
‘Körkort\n’ +
‘Ajokoruli\n’ +
‘Permis de conduire\n’ +
‘4а. 1998-01-12\n’ +
‘Carta de Condução -\n’ +
‘Rifbeteljs\n’ +
‘4b. -\n’ +
‘4c. STATENS VEGVESEN\n’ +
‘Permiso de Conducción\n’ +
‘Filhirerschews\n’ +
‘4d. 020759 23456\n’ +
‘Patente di guida\n’ +
‘5.\n’ +
‘67 98 123456 1\n’ +
‘Drivig Licence\n’ +
‘dinas Tomina\n’ +
‘Калі\n’ +
‘9. A1 BE CIE C\n’ +
‘Nardmann\n’ +
‘Okuskirteint\n’ +
‘EOS-modell’,
ErrorMessage: ‘’,
ErrorDetails: ‘’
}
],

I confirmed the issue. As workaround I suggest that you use the table ocr feature. This way each item is in its own line. This can be used to identify it.

1) WITHOUT table ocr (isTable = false)

N
FORERKORT
NORGE
1. NORDMANN
Koykor:
KARI
AjokorEi
3. 1959-07-02-
Permis de condaire
4a.1998-01-12
Abeua Oomons
Surla ae Condugao
Rijbevoiis
40.

2) WITH table ocr (isTable = true)

N	FORERKORT	NORGE	
1. NORDMANN	Koykor:	
KARI	AjokorEi	
3. 1959-07-02-	Permis de condaire	
Abeua Oomons	
4a.1998-01-12	
40.	Surla ae Condugao	Rijbevoiis

I have tried with the isTable but i still cannot get it to work properly. Could the pro tier with 5mb upload limit maybe improve the quality of the images and make the OCR better? Do you have access to try it with the pro tier?

Greetings,
Kuken666

Sure, I can test this for you. Can you upload a higher quality image to this post?

Thanks, but i can’t find any high resolution example id-card anywhere, and i am not allowed to upload a real license either. Not sure what to do :smiling_face_with_tear: