The Azure Document Translation service can sometimes face challenges with accuracy, especially when translating PDF files.
- Text in Images: The Azure translation service does not translate text that is embedded in images within a document. If the highlighted areas you mentioned contain text in images, this would explain why they are not translated.
- Document Format: The quality of translation can vary depending on the format of the PDF. Native PDFs (those generated from digital file formats) generally provide better results compared to scanned PDFs, which may lose formatting and layout during translation.
- Mixed Language Input: If your PDF contains text in multiple languages, the translation might not be optimal, leading to some text being left untranslated or mistranslated.
- Character Count and Context: Machine translation systems often translate documents sentence by sentence without full context, which can lead to inaccuracies, especially with pronouns or gendered language.
Kindly refer below link: requirements-and-limitations
use-document-translation-sdk-in-your-applications
Thank You.