Learning how to build a human review workflow
Before you get started with creating a human review workflow with Amazon A2I, you will need an S3 bucket in the same AWS Region to collect output data for the workflow.
Creating a labeling workforce
To build a human review workflow, you will need to define the labeling workforce who will label your dataset. As we learned in Chapter 9, you can choose either a public (Mechanical Turk), private (in-house), or vendor (AWS Marketplace/third-party) workforce to review your dataset. You can create and manage workforces from Amazon Ground Truth’s console. If you have an existing workforce in Ground Truth, you can use the same workforce to review predictions.
Setting up an A2I human review workflow or flow definition
As we learned in the Core concepts of Amazon A2I section, next, we’ll create a human review workflow to define the task types, the activation conditions if you’re using a built-in task type, task template...