Perception Systems: How Robots and Cars See, Fuse and Track the World › Module

3D Object Detection: Boxes, Classes, and an Honest Score

Read a 3D box, explain exactly why class-agnostic mAP is 0.383 but six-class mAP is only 0.080, and see what pillar, voxel and camera-only 3D detectors change.

Course access required · Part of Perception Systems: How Robots and Cars See, Fuse and Track the World

Open module