Amazon Rekognition is a service provided by AWS that uses deep learning technology to analyze images and videos for various applications.

How Amazon Rekognition Works

• Amazon Rekognition provides two API sets: Amazon Rekognition Image for analyzing images, and Amazon Rekognition Video for analyzing videos1.
• Both APIs analyze images and videos to provide insights that can be used in applications1.
• For example, when a customer uploads a photo, Amazon Rekognition Image can detect real-world objects or faces in the image1.
• Amazon Rekognition Video can track the path of people in a stored video or search a streaming video for persons whose facial descriptions match facial descriptions already stored by Amazon Rekognition1.

Benefits of Amazon Rekognition

• Amazon Rekognition makes it easy to add powerful visual analysis to your applications2.
• It provides highly accurate facial analysis, face comparison, and face search capabilities2.
• It allows you to enhance the accuracy of its certain deep learning models by training a custom adapter1.

Limitations of Amazon Rekognition

• The maximum image size stored as an Amazon S3 object is limited to 15 MB5.
• The maximum image dimensions for DetectModerationLabels, DetectLabels is 10K pixels for both width and height5.
• The maximum images size as raw bytes passed in as a parameter to an API is 5 MB5.
• Amazon Rekognition supports only the PNG and JPEG image formats5.

Use Cases of Amazon Rekognition

• Amazon Rekognition can be used in various domains such as public safety and law enforcement scenarios—such as finding lost children, combating human trafficking, or preventing crimes6.
• It can be used to monitor business performance, identify unusual variances in subscriptions, conversion rates, and revenue7.
• It can be used to improve customer experience by detecting metric spikes and dips to better understand customer-related issues, churn rates, and install or purchase rates7.
• It can be used to optimize digital ad spend by automatically understanding when your campaign is overspending, underperforming, or encountering errors, without the need for manual intervention7.