Hello. I have noticed while using the ROS2 wrapper that the x position for human detection is sometimes a negative value. Should this be possible? The values are also sometimes much higher than expected. I am using a ZED2i with the jetson orin nano.
Hi @markaajan
if the tracking is active it’s possible that the prediction step of the Kalman filter predicted such a position according to the object speed when it was not more detected.
I disabled SDK predictions with the prediction timeout parameter and it doesn’t seem to have negative values anymore. However, I am getting nan values sometimes for the detection. I have attached the config file I am using.