![background image](http://html1.mh-extra.com/html/ganz/zn-aibox-std-pro/zn-aibox-std-pro_user-manual_4324350032.webp)
32
Reduce False Detection Setting
Deep learning object detection cannot be 100% accurate.
There are several tools to reduce false detections and false alarms.
Learn more about these features below and add settings to reduce false detection.
•
Object Size Filter
•
Object Exclusion Area
1. Object Size Filter
Within the same field of view, the size of objects of the same type will be approximately
constant, or if the field of view is narrow and the distance is close, the size of objects at the top
and bottom will increase and decrease at a constant rate and be detected.
These characteristics can be used to exclude detected objects from events if their size is too
large or small compared to expectations.
1.1 Object Minimum Size Filter
The Object Minimum Size Filter is a setting that allows a detected object to be recognized as an
object only if the size of its bounding box is greater than the size of the box you set.
To access the settings, click Object Size Filter in the sidebar menu and select Min Size Filter in
the body area.
1.1.1 How to Filter The Minimum Object Size
If the bounding box of an object is even larger by one horizontal or vertical dimension than the
minimum size filter of the object, it will not be filtered out. Only when the object’s bounding box
is completely within the minimum size filter will the object be filtered out. See the illustration
above to see how the minimum size filter works and which objects are filtered based on the
object’s bounding box size.