Main Content

View Summary of Ground Truth Labels

You can use the Image Labeler, Video Labeler, and Ground Truth Labeler (Automated Driving Toolbox) (requires Automated Driving Toolbox™) apps to interactively label ground truth data in image collections, videos, image sequences, or lidar point clouds. For details about the supported data sources, see Choose an App to Label Ground Truth Data.

You can view and compare the distribution of ROI and scene labels by clicking View Label Summary on the app toolstrip.

View Label Summary

Clicking View Label Summary opens dockable distribution graphs for the ROI and scene labels.

The x-axis of the graph displays the timestamps across the duration of the video, image sequence, or lidar signal. Units are in seconds. For image collections (Image Labeler app only), the x-axis displays the numeric ID of each image in the collection.

For all ROI labels except pixels, the y-axis displays the number of ROIs at each timestamp or for each image. The visual summary does not include information about sublabels or label attributes.

For pixel ROI labels, the y-axis displays the percentage of the frame that is labeled for each pixel label.

For scene labels, the graph displays the presence or absence of a scene label at each timestamp or for each image in a collection.

In the Ground Truth Labeler app, you can view labels by signal. From Signal Name, select a signal to view a summary of the labels for that signal.

Use the graphs to examine the occurrence of labels over time or in relation to each other. Drag the black vertical line in any graph to move to a different timestamp or image in a collection.

To dock the Label Summary window in your workspace, select Dock Label Summaryfrom the top-right corner of Label Summary window.

Compare Selected Labels

To selectively compare labels, select specific label check boxes and then click Compare Selected Labels. The Label Summary window displays ROI labels selected for comparison on a single graph.

See Also

Apps

Objects

Related Topics