A computer vision startup needs thousands of images annotated with bounding boxes. It wants a managed labeling service that can draw on its own staff, a vendor team, or a public crowd, and that trains a model to auto-label the easy images so people focus on the hard ones. Which service should it use?
Show answer and explanation
Answer: C
Ground Truth coordinates annotation work across private teams, vendor workforces, or public crowds. Its automated labeling uses active learning: a model labels the examples it is confident about and humans handle the uncertain remainder, which lowers labeling cost. Canvas is a no-code modeling tool, DataBrew cleans and transforms tabular data, and Data Wrangler prepares data for ML inside Studio. None of those three manages annotation workforces.
Why the other options are wrong
- A. SageMaker Canvas lets business users build predictive models without code. It trains models on labeled data but does not orchestrate human annotators or bounding-box labeling tasks.
- B. AWS Glue DataBrew is a visual tool for cleaning and transforming tabular datasets. It does not handle image annotation or route work to human labelers.
- D. Data Wrangler offers visual data preparation and feature transformations inside SageMaker Studio. It does not provide annotation workforces or automated labeling for images.