Problem Description
In scenes with partial or full occlusions, the localization point derived from the lower midpoint of the detected bounding box can become inaccurate. This may result in tracked objects being positioned in physically impossible locations (e.g., inside walls, rooms, or restricted areas), reducing tracking reliability.
Proposed Solution
Introduce configurable Ignore Zones / Invalid Localization Regions that allow users to define areas affected by persistent occlusions or geometric constraints. Any localization coordinates falling within these regions can be ignored, filtered, or corrected before being used by the tracking engine.
Examples
- A person cannot be inside a wall.
- A forklift cannot appear inside a building room if it is supposed to be moving on a particular path.
- A tracked object cannot suddenly jump across a barrier.
- A person can only be localized within walkable areas.
Problem Description
In scenes with partial or full occlusions, the localization point derived from the lower midpoint of the detected bounding box can become inaccurate. This may result in tracked objects being positioned in physically impossible locations (e.g., inside walls, rooms, or restricted areas), reducing tracking reliability.
Proposed Solution
Introduce configurable Ignore Zones / Invalid Localization Regions that allow users to define areas affected by persistent occlusions or geometric constraints. Any localization coordinates falling within these regions can be ignored, filtered, or corrected before being used by the tracking engine.
Examples