A company wants to detect which houses have solar panels from satellite imagery. They collected 8,000 training images and will use SageMaker Ground Truth for labeling. The internal team is small and has no ML experience. Which workflow requires the least effort from this team?

The correct option leverages Amazon Rekognition Custom Labels, which is a fully managed service that allows users to train custom object detection models without writing any code or having ML expertise. This aligns with the requirement of minimal effort and no ML experience. Ground Truth's active learning feature further reduces labeling effort by intelligently selecting the most informative images for human review, thus optimizing the small internal team's time. The incorrect options either involve manual labeling (which is more effort than active learning), or require using SageMaker's built-in object detection algorithm. While powerful, SageMaker's algorithms typically demand more ML expertise for training, hyperparameter tuning, and deployment compared to the fully managed Rekognition Custom Labels. Using a public workforce might reduce internal effort for labeling, but the subsequent use of SageMaker's built-in algorithm still introduces the need for ML expertise, which the team lacks.