Textract query result removing characters

The issue you're experiencing with Amazon Textract removing characters, specifically the '<' symbols, from your MRZ (Machine Readable Zone) line is a known limitation of the service when dealing with certain special characters.

Textract can sometimes have difficulties accurately processing or preserving special characters, especially in custom query results. This behavior is not uncommon, particularly with characters that may have special meaning in certain contexts, like '<' which could be interpreted as a formatting or markup character in some systems.

To address this issue, you have a few potential options:

1. Post-processing: Implement a post-processing step in your code to add back the missing characters. Since MRZ lines have a standard format, you could potentially reconstruct the full line based on the expected length and pattern.

2. Use raw OCR results: Instead of relying solely on the custom query result, you might want to cross-reference with the raw OCR output from Textract. The full text, including special characters, might be preserved in the general OCR results.

3. Image preprocessing: Ensure that the image quality is as high as possible. Clear, well-contrasted images can improve Textract's ability to accurately recognize all characters.

4. Feedback to AWS: Consider providing feedback to AWS about this specific issue. While Textract is not directly trainable by users, AWS does use customer feedback to improve the service over time.

5. Alternative extraction method: If this issue persists and is critical for your use case, you might need to explore alternative methods or services for extracting MRZ data that can handle these special characters more reliably.

Remember, while Textract is powerful for many document processing tasks, it may require additional handling for specific use cases like MRZ reading where every character, including special characters, is crucial.
Sources
Textract errors | AWS re:Post
Amazon Textract Training | AWS re:Post