FAQ

I already have image classification models containerized and deployed in AWS Fargate. What is the advantage of moving to an Amazon SageMaker AI serverless deployment?

SageMaker AI offers tools for model training, monitoring, and deployment, which work within a standardized API. If you do not plan to make use of these features, there might not be a reason to change your deployment strategy.

How can I incorporate a managed annotation solution into a retraining workflow?

Amazon SageMaker Ground Truth provides an annotation solution for image classification that integrates with the rest of the SageMaker AI services. For more information, see Image Classification (Single Label) and Image Classification (Multi-label) in the SageMaker AI Developer Guide.

How can I make sure my image classification model is fair and accurate?

You can use services, such as Amazon SageMaker AI Clarify, to detect potential bias. You can also implement model monitoring and continuous evaluation with Amazon SageMaker AI Model Monitor. We recommend that you follow AWS guidance for responsible AI and use Amazon SageMaker Ground Truth to create high-quality training data. We also recommend that you regularly retrain and update your model with new and diverse data.

Can I use my own pretrained image classification model with Amazon Rekognition or Amazon Rekognition Custom Labels?

No, Amazon Rekognition and Amazon Rekognition Custom Labels do not allow you to use your own pretrained models. You can deploy your existing pretrained model by using Amazon SageMaker AI or a custom container solution on Amazon Elastic Container Service (Amazon ECS) or Amazon Elastic Kubernetes Service (Amazon EKS).