Learning 3D Information from 2D Images Using Aperture Rendering Generative Adversarial Networks toward Developing a Computer that “Understands the 3D World”

When people look at photos, they can estimate three-dimensional (3D) information, such as depth, from their experience and knowledge, but computers have difficulty in doing so because they cannot have such experience and knowledge. We spoke to Takuhiro Kaneko, a distinguished researcher who developed a novel deep learning model that can learn 3D information from standard 2D images such as those on the web. © 2022 Authors. All rights reserved.