A FiftyOne panel plugin that displays selected samples in a detail view with segmentation mask overlays, image adjustment controls, and class filtering.
Select one or more samples in the grid and the panel renders them with interactive controls for inspecting segmentation masks (both semantic and instance).
fiftyone plugins download https://github.com/allenleetc/sample-inspector- Multi-sample grid — select multiple samples and view them side by side
- Semantic segmentation overlays — renders
mask_pathfields with color-coded class overlays - Instance segmentation overlays — renders
Detectionswith per-instance mask overlays positioned at bounding boxes - Brightness / Contrast — adjustable sliders (0-200%)
- Overlay opacity — control mask transparency (0-100%)
- Class filtering — toggle individual classes on/off with chip selectors
An unlisted Python operator that fetches mask_targets and default_mask_targets from the dataset. Called internally by the panel to map pixel values to class labels for semantic segmentation overlays.
- Open the FiftyOne App and load a dataset with segmentation labels
- Open the Sample Inspector panel from the panels menu
- Select one or more samples in the grid
- Use the sliders to adjust brightness, contrast, and overlay opacity
- Use the class filter chips to toggle specific classes on/off
Requires a local clone of the FiftyOne repository and FIFTYONE_DIR set to its root.
yarn install
yarn buildSee FiftyOne JS Plugin Build Utils for build configuration details.
