Is there anything I can do to improve the OCR quality of the Cloud SDK? The output of FineReader 10 is more consistent.
For one particular document (I can share if required), the first line OCR is different
Original: 2 . Afte r you have divided your paper into fractions, color each fraction a different color.
FineReader 10: 2. After you have divided your paper into fractions, color each fraction a different color.
Cloud SDK: 2 . A fte r you have divided your paper into fractions, color each fraction a different color.
Comments
3 comments
Does not look like problems with character recognition at all, since in both cases text is the same, but spaces positioning is different. How do you get the text and what is your output format? Problems like that might appear doe to character positioning inside PDF, and Adobe may "invent" spaces sometime where they are not present.
Output for Cloud SDK is PDFSearchable and use the processImage Method. PDF is obtained using CamScanner for Android.
Please provide us with the document you process for investigation - send it to cloudocrsdk@abbyy.com.
Please sign in to leave a comment.