bda does not alter the input image or the ground truth depth; it only transforms the box ground truth in bev space. Therefore, it is sufficient to apply the transformation only to the BEV features recovered from LSS. Cloud anyone give me some help abount the reason,Thanks.
bda does not alter the input image or the ground truth depth; it only transforms the box ground truth in bev space. Therefore, it is sufficient to apply the transformation only to the BEV features recovered from LSS. Cloud anyone give me some help abount the reason,Thanks.