Image Generation From Small Datasets via Batch Statistics Adaptation

2019/04/03 by Atsuhiro Noguchi, Noguchi, Atsuhiro, Tatsuya Harada +1
Computer Science · Mathematics · #Advanced Image Processing Techniques #Artificial intelligence #Class (philosophy) #Computer Vision and Pattern Recognition (cs.CV) #Computer science #Data mining #Digital Media Forensic Detection #Domain (mathematical analysis) #Domain adaptation #FOS: Computer and information sciences #Fidelity #Generative Adversarial Networks and Image Synthesis #Generative grammar #Generator (circuit theory) #Image (mathematics) #Machine learning #Mathematics #Pattern recognition (psychology) #Quality (philosophy) #cs.CV

paper · pdf · doi:10.48550/arxiv.1904.01774

ICCV 2019

openalex publication_date 2019/04/03 · arxiv created 2019/10/23 · arxiv updated 2019/10/24 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28

Abstract

Thanks to the recent development of deep generative models, it is becoming possible to generate high-quality images with both fidelity and diversity. However, the training of such generative models requires a large dataset. To reduce the amount of data required, we propose a new method for transferring prior knowledge of the pre-trained generator, which is trained with a large dataset, to a small dataset in a different domain. Using such prior knowledge, the model can generate images leveraging some common sense that cannot be acquired from a small dataset. In this work, we propose a novel method focusing on the parameters for batch statistics, scale and shift, of the hidden layers in the generator. By training only these parameters in a supervised manner, we achieved stable training of the generator, and our method can generate higher quality images compared to previous methods without collapsing, even when the dataset is small (~100). Our results show that the diversity of the filters acquired in the pre-trained generator is important for the performance on the target domain. Our method makes it possible to add a new class or domain to a pre-trained generator without disturbing the performance on the original domain.

Citations

• LSUN: Construction of a Large-scale Image Dataset using Deep Learning with Humans in the Loop
• Batch Normalization: Accelerating Deep Network Training by Reducing Internal Covariate Shift
• Auto-Encoding Variational Bayes
• Semi-Supervised Learning with Deep Generative Models
• ImageNet Large Scale Visual Recognition Challenge
• Learning Transferable Features with Deep Adaptation Networks
• Domain Separation Networks
• A Learned Representation For Artistic Style
• Generative Knowledge Transfer for Neural Language Models
• Transferring GANs: generating images from limited data
• Self-Attention Generative Adversarial Networks
• IntroVAE: Introspective Variational Autoencoders for Photographic Image Synthesis
• Large Scale GAN Training for High Fidelity Natural Image Synthesis
• Generating High Fidelity Images with Subscale Pixel Networks and Multidimensional Upscaling
• A Style-Based Generator Architecture for Generative Adversarial Networks
• Adversarial Discriminative Domain Adaptation
• Unsupervised Representation Learning with Deep Convolutional Generative Adversarial Networks

Related

• Image Synthesis under Limited Data: A Survey and Taxonomy
• This dataset does not exist: training models from generated images
• Unsupervised Domain Adaptation using Generative Models and Self-ensembling
• Batch Normalization in the final layer of generative networks
• Peer is Your Pillar: A Data-unbalanced Conditional GANs for Few-shot Image Generation