IVS Global Setup Failing

[Nd1776]

To take some of the load off BI/CPU, I'm transitioning to the IVS AI in the Dahua cameras. The global IVS setup is proving to be a bit of a problem. Following the Dahua Wiki, I've drawn the 3 vertical and 1 horizontal lines, all the same length. But during the calibration validation, the measurements are way off. What was set with an actual height of 1m is measured at 1.5m for the closer line and about 1/2m for the furthest line. It seems to have no issue with the horizontal length.

Anyone have any pointers for the global setup or any ideas how to resolve it? Thanks! Apologies if this has been answered somewhere else, but during a forum search for "IVS" it somehow returns 0 results.

 

[wittaj]

Was the camera having trouble with the AI?

I have only done the global setup if the camera was false triggering. Most of mine haven't had to do the global config.

Since your global config is way off, try a longer object for the measuring. I found one close, middle and far have resulted in the closest calibration.

 

[looney2ns]

Did you use a known size object to measure on screen?
Example, take a 1 meter board, and place it at three different distances from the camera. At each location, use the GS measurement on each location.

 

[Nd1776]

Was the camera having trouble with the AI?

I have only done the global setup if the camera was false triggering. Most of mine haven't had to do the global config.

Since your global config is way off, try a longer object for the measuring. I found one close, middle and far have resulted in the closest calibration.

Cameras seem to be working pretty well without the global setup, only occasional false triggers are from my human sized dog. I may be shooting myself in the foot by complicating things.

 

[Nd1776]

Did you use a known size object to measure on screen?
Example, take a 1 meter board, and place it at three different distances from the camera. At each location, use the GS measurement on each location.

Did exactly that, a 1 meter marked pole at 3 different distances. All lengths set to exactly one meter, and then the IVS calibration totally balls it up. I may try wittaj's suggestion to use a longer object just to see what that does.