Can Amazon Rekognition do content moderation on PDF files?

Amazon Rekognition itself does not directly perform content moderation on PDF files. Rekognition is primarily designed for image and video analysis, including content moderation for these formats.

However, to moderate content in PDF files, you can use a combination of AWS services:

1. First, you would use Amazon Textract to extract text and data from the PDF document. Textract can handle scanned documents and extract both printed and handwritten text.

2. Once the text is extracted, you can use Amazon Comprehend for text analysis. Comprehend can detect sentiment, key phrases, and personally identifiable information (PII) in the extracted text.

3. If the PDF contains images, you can extract those images and then use Amazon Rekognition to perform content moderation on them. Rekognition can detect inappropriate, unwanted, or offensive content in images.

4. For any text extracted from the PDF (either directly or from images), you could use Amazon Translate if needed to translate the content before analysis.

This combination of services allows you to create a comprehensive content moderation workflow for PDF files, leveraging the strengths of each service to analyze different aspects of the document's content.
Sources
Content moderation design patterns with AWS managed AI services | AWS Machine Learning Blog
Moderating content - Amazon Rekognition