Deep Generative Image Models using a Laplacian Pyramid of Adversarial Networks

Abstract

In this paper we introduce a generative parametric model capable of producing high quality samples of natural images. Our approach uses a cascade of convolutional networks within a Laplacian pyramid framework to generate images in a coarse-to-fine fashion. At each level of the pyramid, a separate generative convnet model is trained using the Generative Adversarial Nets (GAN) approach (Goodfellow et al.). Samples drawn from our model are of significantly higher quality than alternate approaches. In a quantitative assessment by human evaluators, our CIFAR10 samples were mistaken for real images around 40% of the time, compared to 10% for samples drawn from a GAN baseline model. We also show samples from models trained on the higher resolution images of the LSUN scene dataset.

References in corpus (2)

• Conditional Generative Adversarial Nets
• DRAW: A Recurrent Neural Network For Image Generation

Cited by in corpus (113)

• Energy-based Generative Adversarial Network
• Generating Videos with Scene Dynamics
• CyCADA: Cycle-Consistent Adversarial Domain Adaptation
• C-RNN-GAN: Continuous recurrent neural networks with adversarial training
• Generating Adversarial Malware Examples for Black-Box Attacks Based on GAN
• Towards Principled Methods for Training Generative Adversarial Networks
• A Review on Generative Adversarial Networks: Algorithms, Theory, and Applications
• Disentangling factors of variation in deep representations using adversarial training
• Learning What and Where to Draw
• GANS for Sequences of Discrete Elements with the Gumbel-softmax Distribution
• FAMED-Net: A Fast and Accurate Multi-scale End-to-end Dehazing Network
• GraphGAN: Graph Representation Learning with Generative Adversarial Nets
• AttnGAN: Fine-Grained Text to Image Generation with Attentional Generative Adversarial Networks
• A Two-Stage Approach to Few-Shot Learning for Image Recognition
• Dual Discriminator Generative Adversarial Nets
• Learning to Protect Communications with Adversarial Neural Cryptography
• Learning to Generate Images of Outdoor Scenes from Attributes and Semantic Layouts
• Real-Time Adaptive Image Compression
• Age Progression/Regression by Conditional Adversarial Autoencoder
• PixelNet: Representation of the pixels, by the pixels, and for the pixels
• Perceptual Generative Adversarial Networks for Small Object Detection
• Best of Both Worlds: Transferring Knowledge from Discriminative Learning to a Generative Visual Dialog Model
• Deep Probabilistic Programming
• Photographic Image Synthesis with Cascaded Refinement Networks
• Triangle Generative Adversarial Networks
• Data Augmentation in Emotion Classification Using Generative Adversarial Networks
• Semantic Facial Expression Editing using Autoencoded Flow
• Improved Adversarial Systems for 3D Object Generation and Reconstruction
• Improving Variational Autoencoder with Deep Feature Consistent and Generative Adversarial Training
• PixelNet: Towards a General Pixel-level Architecture
• Label Refinement Network for Coarse-to-Fine Semantic Segmentation
• Generative Adversarial Nets from a Density Ratio Estimation Perspective
• A Hybrid Convolutional Variational Autoencoder for Text Generation
• Ensembles of Generative Adversarial Networks
• Early Action Prediction with Generative Adversarial Networks
• Generating Images Part by Part with Composite Generative Adversarial Networks
• Generative Adversarial Parallelization
• Survey on Deep Multi-modal Data Analytics: Collaboration, Rivalry and Fusion
• Video Imagination from a Single Image with Transformation Generation
• Diversified Texture Synthesis with Feed-forward Networks
• Shape Inpainting using 3D Generative Adversarial Network and Recurrent Convolutional Networks
• Regularization Methods for Generative Adversarial Networks: An Overview of Recent Studies
• Comparative Study on Generative Adversarial Networks
• Birds of A Feather Flock Together: Category-Divergence Guidance for Domain Adaptive Segmentation
• FCC-GAN: A Fully Connected and Convolutional Net Architecture for GANs
• High-Quality Face Image SR Using Conditional Generative Adversarial Networks
• Learning Kernel for Conditional Moment-Matching Discrepancy-based Image Classification
• Bandwidth Extension on Raw Audio via Generative Adversarial Networks
• Optical Flow Estimation using a Spatial Pyramid Network
• Synthesizing Filamentary Structured Images with GANs
• 3D Shape Induction from 2D Views of Multiple Objects
• eCommerceGAN : A Generative Adversarial Network for E-commerce
• Identity-preserving Face Recovery from Stylized Portraits
• PTNet: A High-Resolution Infant MRI Synthesizer Based on Transformer
• A Review of Semi Supervised Learning Theories and Recent Advances
• Energy-based Models for Video Anomaly Detection
• Photorealistic Facial Expression Synthesis by the Conditional Difference Adversarial Autoencoder
• Are You Talking to Me? Reasoned Visual Dialog Generation through Adversarial Learning
• Biased Mixtures Of Experts: Enabling Computer Vision Inference Under Data Transfer Limitations
• A Novel Convolutional Neural Network for Image Steganalysis with Shared Normalization
• Sequential IoT Data Augmentation using Generative Adversarial Networks
• DeLiGAN : Generative Adversarial Networks for Diverse and Limited Data
• Neural Networks for Modeling Source Code Edits
• Semantically Consistent Image Completion with Fine-grained Details
• Towards cross-lingual distributed representations without parallel text trained with adversarial autoencoders
• Adversarial Learning for Chinese NER from Crowd Annotations
• DeepWarp: Photorealistic Image Resynthesis for Gaze Manipulation
• GILT: Generating Images from Long Text
• A Survey and Taxonomy of Adversarial Neural Networks for Text-to-Image Synthesis
• Annealed Generative Adversarial Networks
• Associative Adversarial Networks
• Towards Proof Synthesis Guided by Neural Machine Translation for Intuitionistic Propositional Logic
• RankGAN: A Maximum Margin Ranking GAN for Generating Faces
• PixelNN: Example-based Image Synthesis
• Hybrid VAE: Improving Deep Generative Models using Partial Observations
• Stopping GAN Violence: Generative Unadversarial Networks
• Teaching Machines to Converse
• Voice Conversion using Convolutional Neural Networks
• Can GAN Learn Topological Features of a Graph?
• Learning Temporal Transformations From Time-Lapse Videos
• Visual Decoding of Targets During Visual Search From Human Eye Fixations
• Generative Collaborative Networks for Single Image Super-Resolution
• Pipeline Generative Adversarial Networks for Facial Images Generation with Multiple Attributes
• Learning Disentangled Representations of Satellite Image Time Series
• Generative Mixture of Networks
• Image2GIF: Generating Cinemagraphs using Recurrent Deep Q-Networks
• Max-Margin Deep Generative Models for (Semi-)Supervised Learning
• Stable Distribution Alignment Using the Dual of the Adversarial Distance
• How to make a pizza: Learning a compositional layer-based GAN model
• Applying Generative Adversarial Networks to Intelligent Subsurface Imaging and Identification
• Dense 3D Point Cloud Reconstruction Using a Deep Pyramid Network
• Portrait Shadow Manipulation
• GMM-Based Generative Adversarial Encoder Learning
• Omni-GAN: On the Secrets of cGANs and Beyond
• Deep Markov Random Field for Image Modeling
• Regularized Sequential Latent Variable Models with Adversarial Neural Networks
• Learning to Avoid Errors in GANs by Manipulating Input Spaces
• Authoring image decompositions with generative models
• Generative Adversarial Networks with Inverse Transformation Unit
• A Study of Cross-domain Generative Models applied to Cartoon Series
• Image Disguise based on Generative Model
• KGAN: How to Break The Minimax Game in GAN
• NeurInt : Learning to Interpolate through Neural ODEs
• 3D-MOV: Audio-Visual LSTM Autoencoder for 3D Reconstruction of Multiple Objects from Video
• LR-to-HR Face Hallucination with an Adversarial Progressive Attribute-Induced Network
• StackGAN: Facial Image Generation Optimizations
• Enhancing audio quality for expressive Neural Text-to-Speech
• Unsupervised Meta-learning of Figure-Ground Segmentation via Imitating Visual Effects
• Deep Convolutional Generative Adversarial Network Based Food Recognition Using Partially Labeled Data
• Virtual Conditional Generative Adversarial Networks
• Adversarial Code Learning for Image Generation
• Forget the Learning Rate, Decay Loss
• Making a Case for Learning Motion Representations with Phase