S3
Go to S3, find the bucket named customized-sagemaker-image-decision-tree-bucket, and click Delete.
ECR
Go to ECR, find the repository named decision-tree, and click Delete.
SageMaker
Go to SageMaker, navigate to Inference → Endpoints. If any endpoint is in the InService state, select it, click Action, then Delete.
The remaining resources, such as Lambda functions, IAM roles, and policies, do not incur charges, so we can keep them for reference.