Interface MediaSessionDetectionOptions

Multiple detection sources composed into one active semantic frame.

Static detection frames known at session creation time.

Caller-owned source for loading detection frames by media-time range.

Session-owned writable source for streaming detections into cold storage. Prefer this name for new code.

Backward-compatible alias for appendable.

Detection-frame selection options shared by the hot buffer and renderer.

Hot detection-window loading options.

Hold playback until detections cover the frame about to be presented.

Off by default: the session presents a frame the detection window does not cover without annotations and draws them once coverage lands. Enabled, the session treats detection coverage as part of media readiness and reports buffering while it waits. Merges over buffer.playbackGate.

The sustained wait is the renderer holding a decoded sample back, so it lasts the length of playback only when the renderer pulls samples: a URL, a Blob, or any media source without a presented-frame channel. A source that presents its own frames drives the playhead itself, which is what createVideoEngineMediaRendererSource returns and therefore what most video sessions actually run on. There the gate holds the start of playback and nothing after it, so a preview opens covered and a producer already running keeps its own pace.

Clock used by supplied detection frames. Use MediaStart when inference timestamps begin at zero independently of the file's encoded PTS.

Redraw automatically when appended detections cover the displayed time.

At most one refresh is in flight at a time, and appends that land outside the currently displayed interval never force a render. Set it to false to drive every redraw with an explicit session.refresh(). Defaults to true.