Difference when setting as framework"'image-classification" and framework="mxnet' in AWS Sagemaker?

Question

I am using AWS Sagemaker to perform image classification. My understanding is that Sagemaker would use MXNet for image classification by default.(based on the answer Sagemaker multi-model endpoints with unsupported built-in algorithms)

Would someone clarify what difference there is between the following two cases?

Case 1

algorithm_image=image_uris.retrieve(
    region=boto3.Session().region_name,
    framework="image-classification"
)

Case 2

algorithm_image=image_uris.retrieve(
    region=boto3.Session().region_name,
    framework="mxnet"
)

Thanks in advance.

Answer 1

image-classification refers to the SageMaker Algorithm. As such the Algorithm container will be used where you just need to send your data as input: https://docs.aws.amazon.com/sagemaker/latest/dg/image-classification.html

mxnet refers to the MXNet Framework container where you have full flexibility to define your training script: https://sagemaker.readthedocs.io/en/stable/frameworks/mxnet/using_mxnet.html