Code Examples

The following examples demonstrate registering, retrieving, and ingesting your data through the Hyper-Datasets python interface.

Registering your Data

• register_dataset_with_roi.py - Demonstrates creating a new DatasetVersion and adding to it frames, supporting ROI annotations and metadata
• register_dataset_masks.py - Demonstrates creating a new DatasetVersion and adding to it frames containing masks. This example also demonstrates the DatasetVersion-level pixel segmentation masks.

After executing either of these scripts, you can view your DatasetVersion contents and details in the UI.

Using your Data

Dataviews

The dataview_example_framegroup.py and dataview_example_singleframe.py examples demonstrate how to use a DataView to retrieve your data as SingleFrames and FrameGroups as part of a running experiment. This is done by creating a DataView query and then retrieving the corresponding frames.

DataView details are displayed in the UI in an experiment's DATAVIEWS tab.

Data Ingestion

The pytorch_dataset_example.py example demonstrates how to feed your DataViews to an ML framework by creating a DataView query and wrapping it as a PyTorch Dataset.

The pytorch_dataset_example_with_masks.py example demonstrates the additional actions required when your frames contain masks.