While, with images, the Z axis (or Annotations list) is the only area where annotations are listed out, videos also display annotations in the X axis (or video timeline).
The video timeline auto-adjusts where annotations appear to minimise the amount of vertical space need to show all annotations, but has no bearing on the actual layering of polygons within video frames.
This makes the Annotations list the source of truth for where polygon layers sit in relation to one another.
For example, in a single frame, this video contains 3 cars and a streetlight in the Annotations timeline. They are ordered by newest annotation:
While the video timeline has the same annotations laid out a bit differently to use space as efficiently as possible
Updated over 1 year ago