The correct option is Detects objects and scenes in a picture. Amazon Rekognition is an image analysis service that identifies objects and scenes in photos as a primary capability.

Rekognition can detect many visual elements including common objects, scenes, faces, text in images and it can also perform celebrity recognition and content moderation. Those analysis features are why the highlighted option is correct and why Rekognition is used for computer vision tasks rather than image editing.

The option Extracts tables and form fields from scanned documents is incorrect because extracting structured data from documents is the domain of Amazon Textract which is designed for form and table extraction.

The option Resizes images to new dimensions is incorrect because resizing is an image processing operation that must be done by application logic or image processing tools rather than by Rekognition.

The option Compresses image files to reduce storage is incorrect because Rekognition does not perform file compression and storage optimization is handled by storage services and image processing utilities.

Full AWS Practitioner Certification Question