Answer https://stackoverflow.com/a/62174368/8117673
Further question is - will it affect the accuracy of text detection by Amazon Textract?
Do I need to pre-process the image to get better result from Amazon Textract?
Asked
Active
Viewed 710 times
1 Answers
0
I converted PDF to PNG using command pdftoppm. In Python -> subprocess.Popen(['pdftoppm -png Sample.pdf Sample'])
The accuracy of Amazon Textract on PDF file was more than the PNG format. Because PDF is the original document.
Himanshu Gupta
- 57
- 1
- 6
-
What sort of inaccuracies did you see with PNG? Are more blocks detected in PDF? Did you see more inaccuracies in text with PNG files vs PDF? – Zaki Aziz Jun 16 '21 at 20:30
-
1@Zakir the inaccuracy was with text in PNG. – Himanshu Gupta Aug 27 '21 at 08:52