This post is targeted towards those who are conversant with 2d object detection but are curious about the methodologies arround cloud based 3d object detection.
The first task revolves around a review of the lidar points in the cloud as indicated in the KITTI dataset. Subsequently, we define 3d object detection with its common regression and classification loss which forms the basis of the model performance measurement. Furthermore, we address the two categories of lidar 3d object detection where the merit and demerit will be explained.
For a braader understanding of this details we encourage you watch this video
