비지도학습 (34) 썸네일형 리스트형 AdaIN,2017 Arbitrary Style Transfer in Real-time with Adaptive Instance Normalization, Cornell University,2017 Abstract The earlier paper(Gatys et al.) recently introduced a neural algorithm that renders a content image in the style of another image, achieving so-called style transfer. However, their framework requires a slow iterative optimization process, which limits its practical application. Fast appr.. Improved Training of Wasserstein GANs,2017 Improved Training of Wasserstein GANs Abstract Generative Adversarial Networks (GANs) are powerful generative models, but suffer from training instability. The recently proposed Wasserstein GAN (WGAN) makes progress toward stable training of GANs, but sometimes can still generate only poor samples or fail to converge. We find that these problems are often due to the use of weight clipping in WGA.. PGGAN(2019) PROGRESSIVE GROWING OF GANS FOR IMPROVED QUALITY, STABILITY, AND VARIATION(NVIDIA,2019) ABSTRACT We describe a new training methodology for generative adversarial networks. The key idea is to grow both the generator and discriminator progressively: starting from a low resolution, we add new layers that model increasingly fine details as training progresses. This both speeds the training up and g.. [논문]StyleGAN,2019 A Style-Based Generator Architecture for Generative Adversarial Networks 생성적 적대 네트워크를위한 스타일 기반 생성기 아키텍처(NVIDIA) Abstract Style Transfer Paper에서 차용한 GAN을 위한 대체 생성기 아키텍처 새로운 아키텍처는 자동으로 학습되고 감독되지 않은 높은 수준의 속성(high-level attributes) (예 : 사람 얼굴에 대해 학습된 포즈 및 정체성)과 생성된 이미지 (예 : 주근깨, 머리카락)의 확률적 변화(stochastic variation)를 분리하고, 직관적이고 합성의 특정 스케일링제어를 가능하게합니다. 새로운 생성기는 기존의 분포 품질 메트릭 측면에서 최첨단을 개선하고, 더 나은 보간 특.. Analyzing and Improving the Image Quality of StyleGAN, 2020(StyleGAN2) Abstract The style-based GAN architecture (StyleGAN) yields state-of-the-art results in data-driven unconditional generative image modeling. We expose and analyze several of its characteristic artifacts, and propose changes in both model architecture and training methods to address them. In particular, we redesign the generator normalization, revisit progressive growing, and regularize the gener.. A Style-Based Generator Architecture for Generative Adversarial Networks, 2019(버전 1) A Style-Based Generator Architecture for Generative Adversarial Networks 생성적 적대 네트워크를위한 스타일 기반 생성기 아키텍처(NVIDIA) Abstract We propose an alternative generator architecture for generative adversarial networks, borrowing from style transfer literature. The new architecture leads to an automatically learned, unsupervised separation of high-level attributes (e.g., pose and identity when trained on hum.. [논문]DiscoFaceGAN,2020 Disentangled and Controllable Face Image Generation via 3D Imitative-Contrastive Learning 3D 모방-대비 학습을 통한 엉킴 및 제어 가능한 얼굴 이미지 생성(Microsoft Research Asia) Abstract We propose DiscoFaceGAN, an approach for face image generation with DISentangled, precisely-COntrollable latent representations for identity of non-existing people, expression, pose, and illumination. We embed 3D priors into adversarial.. Patch-Based Image Inpainting with Generative Adversarial Networks,2018 Patch-Based Image Inpainting with Generative Adversarial Networks, 2018 Abstract 1. Introduction 2. Related works 3. Proposed Method 3.1 Generator network 3.2 Discriminator network 3.3 Objective function(Reconstruction loss/Adversarial Loss/Joint Loss) 4. Results 4.1 Datasets(Paris Street View/Google Street View/Places) 4.2 Training details and implementation 4.3 Ablation study 4.4 Comparative e.. 이전 1 2 3 4 5 다음
