Editing natural images is time-consuming and tedious, even with modern digital tools. Given a sufficiently expressive image model, we should be able to intuitively edit images by modifying the model's descriptive variables. The Neural Photo Editor is a simple interface that leverages the power of a novel hybrid generative model to allow a user to directly edit existing photos. Central to the operation of the Neural Photo Editor is the Introspective Adversarial Network, a novel hybridization of the Variational Autoencoder and the Generative Adversarial Network which leverages the inference capabilites of the VAE and the output quality of the GAN to produce high quality samples and reconstructions.
Andrew Brock's background is in computational heat transfer and fluid dynamics for hybrid rocketry, control systems engineering for full-body haptic feedback, and social dance instruction. He holds an MSME from Cal Poly SLO.