The Google team behind Objectron, then, developed a toolset that allowed annotators to label 3D bounding boxes (i.e., rectangular borders) for objects using a split-screen view to display 2D video ...

So first his system renders a 3D reconstruction of the 2D image in very low resolution. You can still get a lot from that — for example, that the outer third of the whole volume appears to be empty.

NVIDIA has introduced a new AI model called Neuralangelo that can create 3D replicas of objects from 2D videos, whether they're classic sculptures or run-of-the-mill trucks and buildings.