Existing 3D scene reconstructions require a cumbersome process of precisely measuring physical spaces with LiDAR or 3D scanners, or correcting thousands of photos along with camera pose information. A ...
Abstract: Knowledge distillation (KD) possesses immense potential to accelerate the deep neural networks (DNNs) for LiDAR-based 3D detection. However, in most of prevailing approaches, the suboptimal ...
Some results have been hidden because they may be inaccessible to you
Show inaccessible results
Feedback