Arbitrary Style Transfer in Real-time with Adaptive Instance Normalization
ICCV17 | 180 | Arbitrary Style Transfer in Real-time with Adaptive Instance Normalization
Xun Huang (Cornell), Serge Belongie ()
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 approximations with feed-forward neural networks have been proposed to speed up neural style transfer. Unfortunately, the speed improvement comes at a cost: the network is usually tied to a fixed set of styles and cannot adapt to arbitrary new styles. In this paper, we present a simple yet effective approach that for the first time enables arbitrary style transfer in real-time. At the heart of our method is a novel adaptive instance normalization (AdaIN) layer that aligns the mean and variance of the content features with those of the style features. Our method achieves speed comparable to the fastest existing approach, without the restriction to a pre-defined set of styles. In addition, our approach allows flexible user controls such as content-style trade-off, style interpolation, color & spatial controls, all using a single feed-forward neural network.
Видео Arbitrary Style Transfer in Real-time with Adaptive Instance Normalization канала ComputerVisionFoundation Videos
Xun Huang (Cornell), Serge Belongie ()
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 approximations with feed-forward neural networks have been proposed to speed up neural style transfer. Unfortunately, the speed improvement comes at a cost: the network is usually tied to a fixed set of styles and cannot adapt to arbitrary new styles. In this paper, we present a simple yet effective approach that for the first time enables arbitrary style transfer in real-time. At the heart of our method is a novel adaptive instance normalization (AdaIN) layer that aligns the mean and variance of the content features with those of the style features. Our method achieves speed comparable to the fastest existing approach, without the restriction to a pre-defined set of styles. In addition, our approach allows flexible user controls such as content-style trade-off, style interpolation, color & spatial controls, all using a single feed-forward neural network.
Видео Arbitrary Style Transfer in Real-time with Adaptive Instance Normalization канала ComputerVisionFoundation Videos
Показать
Комментарии отсутствуют
Информация о видео
17 ноября 2017 г. 3:35:16
00:09:55
Другие видео канала
![Batch Normalization (“batch norm”) explained](https://i.ytimg.com/vi/dXB-KQYkzNU/default.jpg)
![Arbitrary Style Transfer in Real Time!](https://i.ytimg.com/vi/jDadRuMSTHQ/default.jpg)
![Style Transfer and Deep Dream: Making Art with AI](https://i.ytimg.com/vi/du_YnoC6dKU/default.jpg)
![Lecture 1 | Introduction to Convolutional Neural Networks for Visual Recognition](https://i.ytimg.com/vi/vT1JzLTH4G4/default.jpg)
![Transformer Neural Networks - EXPLAINED! (Attention is all you need)](https://i.ytimg.com/vi/TQQlZhbC5ps/default.jpg)
![Photo-Realistic Single Image Super-Resolution Using a Generative Adversarial Network](https://i.ytimg.com/vi/BXIR_SVCrsE/default.jpg)
![Variational Autoencoders](https://i.ytimg.com/vi/9zKuYvjFFS8/default.jpg)
![StyleGAN Part 1: Adaptive Instance Normalization - Matchue GANs](https://i.ytimg.com/vi/idIHmtjz1EQ/default.jpg)
![Leon Gatys shares a neural algorithm of artistic style](https://i.ytimg.com/vi/N14_w2RG1A8/default.jpg)
![Is Style Transfer For Fluid Simulations Possible? 🌊](https://i.ytimg.com/vi/b8sCSumMUvM/default.jpg)
![StyleGAN](https://i.ytimg.com/vi/AQBti_wN414/default.jpg)
![GauGAN Explained!](https://i.ytimg.com/vi/dOjES0yJx7s/default.jpg)
![Lecture 12 | Visualizing and Understanding](https://i.ytimg.com/vi/6wcs6szJWMY/default.jpg)
![Evolving Normalization-Activation Layers](https://i.ytimg.com/vi/klPuEHCKG9M/default.jpg)
![Advanced Theory | Neural Style Transfer #4](https://i.ytimg.com/vi/8pp0Oa3t52s/default.jpg)
![How Convolutional Neural Networks work](https://i.ytimg.com/vi/FmpDIaiMIeA/default.jpg)
![Face editing with Generative Adversarial Networks](https://i.ytimg.com/vi/dCKbRCUyop8/default.jpg)
![Why Does Batch Norm Work? (C2W3L06)](https://i.ytimg.com/vi/nUUqwaxLnWs/default.jpg)
![More with Less: Data Augmentation and Gradient Penalties - Matchue GANs](https://i.ytimg.com/vi/weRmcRzMfUQ/default.jpg)
![BERT Neural Network - EXPLAINED!](https://i.ytimg.com/vi/xI0HHN5XKDo/default.jpg)