Hi @weiyithu , thank you for your great work. Your algorithm works very well in KITTI dataset.
However, in your paper, you didn't mention how to estimate the object's orientation (heading-specific). As far as I understand, if we can determine the rotation angle 'phi' of the bounding box in bird-eye-view, the final object's orientation can be 'phi' or 'phi + pi' because we do not know about the heading of the object. How can you resolve this scenario? Thank you.
Hi @weiyithu , thank you for your great work. Your algorithm works very well in KITTI dataset.
However, in your paper, you didn't mention how to estimate the object's orientation (heading-specific). As far as I understand, if we can determine the rotation angle 'phi' of the bounding box in bird-eye-view, the final object's orientation can be 'phi' or 'phi + pi' because we do not know about the heading of the object. How can you resolve this scenario? Thank you.