As in followed link, mmdetection3d uses the negative of yaw when converting to nuscenes-format bbox in camera cooridnate system. Does nuscenes define yaw in camera coordinate system in another fashion?
https://github.com/open-mmlab/mmdetection3d/blob/962fc830daebad3f93fc7cbf970728057c64977b/mmdet3d/datasets/nuscenes_mono_dataset.py#L713
Considering following scenario, an object in front of a camera driving towards, in mmdetection3d the yaw should be negative 0.5pi. Is it negative 0.5pi in nuscenes?
As in followed link, mmdetection3d uses the negative of yaw when converting to nuscenes-format bbox in camera cooridnate system. Does nuscenes define yaw in camera coordinate system in another fashion?
https://github.com/open-mmlab/mmdetection3d/blob/962fc830daebad3f93fc7cbf970728057c64977b/mmdet3d/datasets/nuscenes_mono_dataset.py#L713
Considering following scenario, an object in front of a camera driving towards, in mmdetection3d the yaw should be negative 0.5pi. Is it negative 0.5pi in nuscenes?