Use Canvas Ready To Use Model

Here is how you can use the Object Detection in Image model from SageMaker Canvas as an endpoint:

1. The Object Detection in Image model is a ready-to-use model available in SageMaker Canvas. However, it seems you are not able to access it as an endpoint outside the Canvas dashboard.
2. SageMaker Canvas now supports deploying Foundation Models (FMs) to SageMaker real-time inference endpoints. This allows you to use generative AI capabilities, like the Object Detection in Image model, outside the Canvas workspace.
3. To deploy the Object Detection in Image model as an endpoint: o Log in to the SageMaker Canvas console o Locate the Object Detection in Image model o Select the model and choose to deploy it as a SageMaker endpoint o Configure the endpoint settings, such as the instance type, and deploy the model
4. Once the model is deployed, you can access it programmatically using the SageMaker SDK or AWS CLI.

https://docs.aws.amazon.com/sagemaker/latest/dg/canvas-ready-to-use-models.html https://docs.aws.amazon.com/sagemaker/latest/dg/object-detection.html

Yes, you can deploy the Amazon SageMaker Canvas Ready-to-use model for Object Detection in Images as an endpoint. This allows you to invoke the model for real-time predictions outside of the SageMaker Canvas application. To deploy a Ready-to-use model as an endpoint from SageMaker Canvas:

1. Open the SageMaker Canvas application and navigate to the model you want to deploy (in this case, the "Object Detection in Images" model).
2. Click on the "Deploy Model" button.
3. In the deployment configuration, you can specify the instance type and instance count for the endpoint based on your performance and cost requirements.
4. For the "Object Detection in Images" model, you can leave the default deployment length as "Deploy indefinitely" unless you want to set a specific time period.
5. Review the deployment configuration and click "Deploy Model".

SageMaker Canvas will now create and deploy the model to a real-time inference endpoint. You can view the deployment details, such as the endpoint name and status, within the Canvas application or in the SageMaker console. Once deployed, you can invoke the endpoint programmatically using the AWS SDK or AWS CLI. Here's an example using the AWS CLI to invoke the endpoint with a JPEG image file:

aws sagemaker-runtime invoke-endpoint \
  --endpoint-name your-endpoint-name \
  --content-type "image/jpeg" \
  --accept "application/json" \
  --body fileb://path/to/image.jpg \
  output.json

The response will contain the detected objects, their class indices, confidence scores, and bounding box coordinates in the specified image. Remember that you will be charged for the SageMaker inference instances while the endpoint is running, so be sure to delete the endpoint when you're done using it to avoid incurring further charges.

Useful sources:

1. https://aws.amazon.com/about-aws/whats-new/2023/10/deploy-ml-models-sagemaker-canvas-real-time-endpoints/
2. https://aws.amazon.com/blogs/machine-learning/deploy-ml-models-built-in-amazon-sagemaker-canvas-to-amazon-sagemaker-real-time-endpoints/
3. https://docs.aws.amazon.com/sagemaker/latest/dg/canvas-deploy-model.html

SageMaker Canvas for its "Objects Detection In Image" model uses Amazon Rekognition, which you can use with a simple API. Here is the documentation showing how to call the Rekognition API to help get the same results in your application.