Pixel space that a detection frame's vector geometry was produced in.
Producers that infer on a resized or transcoded copy of the media can attach
this to a DetectionFrame instead of scaling geometry themselves. Masks are
unaffected: they already carry their own intrinsic width/height.
Pixel space that a detection frame's vector geometry was produced in.
Producers that infer on a resized or transcoded copy of the media can attach this to a
DetectionFrameinstead of scaling geometry themselves. Masks are unaffected: they already carry their own intrinsicwidth/height.