Guided Image Generation with Conditional Invertible Neural Networks

Abstract

In this work, we address the task of natural image generation guided by a conditioning input. We introduce a new architecture called conditional invertible neural network (cINN). The cINN combines the purely generative INN model with an unconstrained feed-forward network, which efficiently preprocesses the conditioning input into useful features. All parameters of the cINN are jointly optimized with a stable, maximum likelihood-based training procedure. By construction, the cINN does not experience mode collapse and generates diverse samples, in contrast to e.g. cGANs. At the same time our model produces sharp images since no reconstruction loss is required, in contrast to e.g. VAEs. We demonstrate these properties for the tasks of MNIST digit generation and image colorization. Furthermore, we take advantage of our bi-directional cINN architecture to explore and manipulate emergent properties of the latent space, such as changing the image style in an intuitive way.

References in corpus (6)

• Conditional Generative Adversarial Nets
• GANs Trained by a Two Time-Scale Update Rule Converge to a Local Nash Equilibrium
• NICE: Non-linear Independent Components Estimation
• A Learned Representation For Artistic Style
• CausalGAN: Learning Causal Implicit Generative Models with Adversarial Training
• Comparison of Maximum Likelihood and GAN-based training of Real NVPs

Cited by in corpus (2)

• Disentanglement by Nonlinear ICA with General Incompressible-flow Networks (GIN)
• A Method to Model Conditional Distributions with Normalizing Flows