Controlling 3D Objects in 2D Image Synthesis

Stephan Brehm, Florian Barthel, Rainer Lienhart

Abstract: In this work, we propose a method that enforces explicit control over various attributes during the image generation process in a generative adversarial net. We propose a semi-supervised learning procedure that allows us to use a quantized approximation of object orientation for learning continuous object rotations. As a result, among many other attributes, our proposed method allows us to control object orientation in scenes that are rendered according to our specifications.