ObjectDetectionAgent Class

Runs object detection over passthrough frames on-device or via a provider, aggregates post-processing, and raises events with decoded boxes for visualization or game logic.

Use this component when your goal is a task-oriented “detect → filter → render/use” loop rather than low-level model handling.

Typical setup: wire a provider such as UnityInferenceEngineProvider and ObjectDetectionVisualizer to draw quads/labels. Frames usually come from the passthrough camera; see the Passthrough overview (guide) and environment depth (guide) to understand how 2D boxes are projected into world space. The agent timeslices inference if supported and exposes an event so UX can react the moment detections are ready.