Abstract: Monocular 3D object detection is a promising yet ill-posed task for autonomous vehicles due to the lack of accurate depth information. Cross-modality knowledge distillation could effectively ...

Abstract: The key to 3D object detection is proper utilization of depth data. Compared with LiDAR based approaches, 3D object detection from a single image remains a challenging task due to the lack ...