Stanford and Google AI researchers introduce MELON: an AI technique that can determine object-centered camera poses completely from scratch while reconstructing the object in 3D
While humans can easily infer the shape of an object from 2D images, computers have difficulty reconstructing accurate 3D models ...