hello author
When the chair object is detected, but the position of the blue three-dimensional frame displayed in rviz is different in semantic segmentation, what should be done?