Find, inspect, and edit dataset samples

Choose All, Training, Validation, or Test and note that later filtering applies within the active scope.

Use cards for visual scanning and quick inspection of image-like or media samples.

Use the compact list when names, states, and metadata matter more than large previews.

Use the table for structured fields and tabular datasets when the option is available.

Check that the displayed count and visible samples match the selected split before taking a bulk action.

Choose the split first so the query does not accidentally mix training and evaluation samples.

Enter a filename, sample name, or other supported query and wait for the result list to update.

Add one label, annotation, state, or other available filter at a time and observe how the result count changes.

Before selecting all or cloning filtered samples, confirm the split, query, active filter chips, and current result count.

Clear the query and filters in reverse order until the expected samples return.

Select a sample and compare its preview, name, metadata, and processing state with the source data.

Inspect available detections, segmentations, and classifications, including label names and values.

Use Previous and Next to review neighboring samples without losing the active result scope.

Wait when a sample is still processing. Record the visible failure when it reaches a failed or unavailable state instead of assuming it is empty.

Open annotation editing or another available sample action only after confirming the sample identity and permissions.

Select individual samples or the visible filtered scope only after checking the result count.

Use the offered edit, annotation, split, copy, remove, delete, or other action for the current data type.

For removal or deletion, read the confirmation and distinguish removing a sample from the dataset from deleting data when the UI presents that choice.

Keep the operation status visible through queued, processing, completed, or failed states.

After completion, refresh the sample count and review failures before repeating an action.

Navigate to the relevant rows and columns, then enter the available edit mode.

Edit supported cells or fields and preserve types expected by the dataset and labels.

Use Remove Row or Delete Row only after reading the displayed scope and consequence.

Delete Column affects the field across the table; verify the column name and downstream task before confirming.

Save the edit, refresh the table, and verify representative rows plus readiness or analysis affected by the change.