To meet these challenges, we present a new approach to model stealing defenses called gradient redirection.
In this paper, we show that respecting equivariance -- the color of a restored pixel should be the same, however the image is cropped -- produces real improvements over the state of the art for restoration.
Recently, StyleGAN has enabled various image manipulation and editing tasks thanks to the high-quality generation and the disentangled latent space.
In this work, we demonstrate the effectiveness of Firth bias reduction in few-shot classification.
While noise inputs to StyleGAN2 are essential for good synthesis, we find that coarse-scale noise interferes with latent variables on this dataset because both control long-scale image effects.
Generative models for 3D shapes represented by hierarchies of parts can generate realistic and diverse sets of outputs.
We present Retrieve in Style (RIS), an unsupervised framework for facial feature transfer and retrieval on real images.
This adversarial loss guarantees the map is diverse -- a very wide range of anime can be produced from a single content code.
Ranked #1 on Image-to-Image Translation on selfie2anime
However, obtaining a realistic image is challenging because the kinematics of garments is complex and because outline, texture, and shading cues in the image reveal errors to human viewers.
Interestingly, the predictions by this model on images with no humans, are also visibly different from the one trained on gendered captions.
Style transfer methods produce a transferred image which is a rendering of a content image in the manner of a style image.
We describe a method that predicts, from a single RGB image, a depth map that describes the scene when a masked object is removed - we call this "counterfactual depth" that models hidden scene geometry together with the observations.
Explaining a deep learning model can help users understand its behavior and allow researchers to discern its shortcomings.
Generative adversarial nets (GANs) and variational auto-encoders have significantly improved our distribution modeling capabilities, showing promise for dataset augmentation, image-to-image translation and feature learning.
We present an object relighting system that allows an artist to select an object from an image and insert it into a target scene.
Outfits in online fashion data are composed of items of many different types (e. g. top, bottom, shoes) that share some stylistic relationship with one another.
Finally, an adversarial pattern on a physical object that could fool a detector would have to be adversarial in the face of a wide family of parametric distortions (scale; view angle; box shift inside the detector; illumination; and so on).
Instead, a trained neural network classifies most of the pictures taken from different distances and angles of a perturbed image correctly.
Instead, we can learn to decompose an image into layers that are "like this" by authoring generative models for each layer using proxy examples that capture the Platonic ideal (Mondrian images for albedo; rendered 3D primitives for shading; material swatches for shading detail).
Such a model is difficult to train, because we do not usually have training data containing many different shadings for the same image.
Our procedure allows speed and accuracy to be traded off in two ways: by choosing the number of Vector Quantization levels, and by choosing to rescore windows or not.