Amazon SageMaker Data Labeling

Amazon SageMaker data labeling offerings provide ML labeling techniques that are less prone to manual errors, including synthetic data generation, and help improve the quality of training datasets. Amazon SageMaker Ground Truth Plus has a multi-step labeling workflow that includes ML models for pre-labeling, machine validation of human labeling to detect errors and low-quality labels, and assistive labeling features (e.g., 3D cuboid snapping, predict-next in video labeling, and auto-segment tools). If you are managing your own data labeling workflows, SageMaker Ground Truth provides automated labeling features such as auto-segment, automatic 3D cuboid snapping, and sensor fusion with 2D video frames. In addition, SageMaker Ground Truth provides automatic data labeling which uses active learning and only routes labeled data to humans if the model cannot confidently label it.

With Amazon SageMaker data labeling offerings, you have two options to label data. First, with SageMaker Ground Truth Plus, an expert workforce that is trained on ML tasks labels your data in accordance with your quality and turnaround time requirements. Second, with SageMaker Ground Truth, you can build and manage your data labeling workflows. You have options to work with labelers inside and outside your organization. For example, you can send labeling jobs to your own labelers, or you can access a workforce of over 500,000 independent contractors who are already performing ML-related tasks through Amazon Mechanical Turk. If your data requires confidentiality or special skills, you can also use vendors that are pre-screened by AWS for quality and security procedures. If you need access to synthetic data so your training datasets are more complete for training ML models, AWS digital artists use customer-provided assets and images to generate synthetic data that is automatically labeled on your behalf.

Amazon SageMaker data labeling offerings enable you to gain transparency into data labeling operations and quality management so you can verify that your quality requirements are being met. SageMaker Ground Truth Plus provides interactive dashboards and user interfaces, so you can monitor progress of training datasets across multiple projects, track project metrics such as daily throughput, inspect labels for quality, and provide feedback on the labeled data.

With Amazon SageMaker data labeling offerings, you can receive high-quality labeled data quickly. With SageMaker Ground Truth Plus, you upload your data in Amazon S3 along with security, privacy, and compliance requirements. AWS experts will then set up the data labeling workflow and an expert workforce will complete your labeling tasks. If you need access to synthetic data, you specify your image requirements or provide 3D assets and baseline images, and SageMaker Ground Truth can generate highly accurate labeled synthetic data for ML model training.