Question

Why does the calculated characters' confidence remain 100% all the time even though the recognition results are marked red?

Answer

This issue is related to the recognition of the PDF text layer.

When a document contains text layer, and Auto or Prefer PDF text layer recognition modes are being used, the confidence of characters is automatically set to 100% since the text layer is considered a trusted resource.

If the OCR Only recognition mode is used, then the confidence is calculated each time when the recognition is performed, and, therefore, its value can be returned by the script.

Therefore, in order to get the characters' confidence, it is recommended to use the OCR Only recognition mode.