You can draw a rectangular selection on the plot to isolate and explore a specific subset of data points. The grid updates instantly to reflect your selection. Use the Order by drop-down to switch the visualization between data and label embeddings.
Prerequisites
- Data Curation: To use embedding plots for data curation, you must upgrade your folder to calculate data embeddings.
- Label Validation: To use embedding plots for label validation, you must calculate metrics and embeddings.
Calculate Metrics & Embeddings
- Navigate to Projects.
- Open the Explore tab.
- Click Metrics & Embeddings.
- Click Compute for either option.
-
Specify the following:
- Similarity & Natural language search and quality metrics: Enable to compute embeddings and quality metrics. Access quality metrics for filtering and sorting.
- Select embeddings: Default embeddings are computed by Encord. Alternatively, import and select your own custom embeddings.
- Embeddings plot, Diversity and Uniqueness metrics: Enable to compute UMAP reduction to generate 2D embeddings plots to visualize your data. Also access diversity and uniqueness metrics for curation.
- Click Start computation.
Using Embedding Plots
- Navigate to Data > Explore.
- Click Frames or Labels.
- Click the Embeddings icon in the toolbar row to open the Embeddings panel as a right-hand panel alongside the grid.
Selecting Data with the Embeddings Panel
Draw a box on the scatter plot to select a subset of data points. The grid on the left immediately filters to show only the items within your selection. Use the unified selection toolbar at the bottom of the grid to add or remove selected items from a collection.Hovering over a point on the scatter plot no longer shows an image preview tooltip. The always-visible grid serves this purpose — items matching your selection appear there instantly.
Label Embedding Plot

