This documentation is for the upcoming version of the AI Manager
Read the documentation for the current version
Ctrlk
AI ManagerNx Meta
For the complete documentation index, see llms.txt. This page is also available as Markdown.

Exclusive Masks

Limit inference to outside a drawn polygon.

Exclusive input masks ignore objects that are fully inside the drawn mask area. Create these masks on the video frame inside the AI Manager's Model Settings, in the Input Masks section.

Select Exclusive from the Mask Type dropdown:

Model Settings panel with the Input Mask Type dropdown open, showing Inclusive and Exclusive options
Choosing the Exclusive mask type

Choosing Exclusive leaves the shape field empty until you draw a mask:

Model Settings panel with Mask Type set to Exclusive and no shape drawn yet
Exclusive mask type selected, no shape drawn yet

Click inside the mask's shape field to open the figure editor and draw the mask over the frame:

Figure editor dialog with a pink rectangle drawn over part of a crowded corridor camera image, resize handles visible
Drawing an exclusive mask shape in the figure editor

Click OK to save the shape. A thumbnail preview appears in the mask's shape field, with a Display on video checkbox to show the mask as an overlay in Live View. Add more masks with the Add button.

Model Settings panel showing an Exclusive input mask thumbnail preview with a pink polygon drawn over part of the camera image
Input Masks setting with an exclusive input mask

Confirm it in Live View: detections inside the mask area are ignored, detections elsewhere are still returned.

Live View of a crowded corridor showing a pink exclusive mask overlay over part of the frame, with detections still boxed in the rest of the frame
Exclusive mask overlay in Live View

In practice, results overlap the polygon borders slightly because the anchor location for a detected object is the middle of the bottom of the detection box.

The center of the bottom of the bounding box is chosen for the anchor location because that is usually a good indication for where a person or object is standing on the ground. In a line-crossing scenario for example, this would make it easier to detect if a person crosses a certain line.

Using an input mask and a region of interest at the same time is not supported.

Click Apply for changes to take effect.

PreviousInclusive Masks
NextClass Visualization

Last updated