GQNs are designed to teach machines a sense of imagination. They take a scene as an input, rotate it, and then predict what it will look like from a different angle – learning to perceive the world in 3D. 27.07.2023 17:54 aior