Regression: Tesseract 5.5.3 recognizes uppercase 'I' as '|' while 5.5.0 recognizes it correctly

Current Behavior

After upgrading Tesseract OCR from 5.5.0 to 5.5.3, OCR incorrectly recognizes the uppercase letter 'I' as the pipe character '|'.

This issue is reproducible with the same:

• Input image
• OCR configuration
• Language (eng)
• Python code
• pytesseract version

Only the Tesseract OCR engine version changed.

Example:

Expected: I have checked all six sides of the package.

Actual: | have checked all six sides of the package.

Configuration:

• Tesseract 5.5.0 -> Correct output
• Tesseract 5.5.3 -> Incorrect output
• OEM: 3
• PSM: 6
• Language: eng

Expected Behavior

Tesseract 5.5.3 should recognize the first character as the uppercase letter 'I', matching the behavior of Tesseract 5.5.0.

Expected output:

I have checked all six sides of the package against Poster 298 and confirmed the customer obliterated any inapplicable markings.

Suggested Fix

The issue appears to be a regression introduced after Tesseract 5.5.0.

Please investigate the OCR behavior changes between versions 5.5.0 and 5.5.3 that affect the recognition of the uppercase letter 'I'.

The same image, OCR configuration, and language data produce the following results:

• Tesseract 5.5.0: Correctly recognizes 'I'
• Tesseract 5.5.3: Recognizes 'I' as '|'

If this behavior is intentional or caused by changes in the OCR engine or trained data, any guidance or workaround would be appreciated.

tesseract -v

5.5.3.20260724

Operating System

No response

Other Operating System

No response

uname -a

No response

Compiler

No response

CPU

No response

Virtualization / Containers

No response

Other Information

No response