For the complete documentation index, see llms.txt. This page is also available as Markdown.

Annotate Video Segments

Mark the start and end of actions in a video.

A segment labels a time range in a video with an action class. In an Action Recognition Project, use segments to show when each action happens.

For example, label opening-box from the first movement of the lid until the box is open. Leave out the time spent walking toward the box or standing beside it afterward.

When you start annotating a video in an Action Recognition Project, you see this annotation editor.

Action Recognition video annotation editor with a video and timeline
The editor shows the video, timeline, action classes, and save status.
1

Add a segment

Drag across the row labeled "Click or drag here to start a segment in a new class" from the start of the action to its end. Select an existing class or create opening-box in the class field that opens.

The segment appears as a colored bar in that class's row. Changes save automatically. Wait for "Saved" in the header before leaving the video.

Action Recognition video annotation editor with a preparing segment on the timeline
Drag across the timeline to add a segment and assign its class.
2

Adjust the start and end

Select the segment and play through the action. Drag either edge of the bar to adjust its time range. For frame-by-frame adjustments, move the playhead to the desired frame, right-click the segment, and choose "Set In" for the start or "Set Out" for the end.

For opening-box, check that the segment begins when the lid starts moving and ends when it is open. Wait for "Saved" after your edits.

Action Recognition video annotation editor with preparing and packaging segments on separate timeline rows
Each action class has its own row, so a video can contain more than one segment.

Last updated

Was this helpful?