Meta Networks for Neural Style Transfer

Abstract

In this paper we propose a new method to get the specified network parameters through one time feed-forward propagation of the meta networks and explore the application to neural style transfer. Recent works on style transfer typically need to train image transformation networks for every new style, and the style is encoded in the network parameters by enormous iterations of stochastic gradient descent. To tackle these issues, we build a meta network which takes in the style image and produces a corresponding image transformations network directly. Compared with optimization-based methods for every style, our meta networks can handle an arbitrary new style within $19ms$ seconds on one modern GPU card. The fast image transformation network generated by our meta network is only 449KB, which is capable of real-time executing on a mobile device. We also investigate the manifold of the style transfer networks by operating the hidden features from meta networks. Experiments have well validated the effectiveness of our method. Code and trained models has been released https://github.com/FalongShen/styletransfer.

References in corpus (10)

• Adam: A Method for Stochastic Optimization
• Very Deep Convolutional Networks for Large-Scale Image Recognition
• Texture Networks: Feed-forward Synthesis of Textures and Stylized Images
• Fast Patch-based Style Transfer of Arbitrary Style
• Stable and Controllable Neural Texture Synthesis and Style Transfer Using Histogram Losses
• Style-Transfer via Texture-Synthesis
• Multi-style Generative Network for Real-time Transfer
• Demystifying Neural Style Transfer
• Diversified Texture Synthesis with Feed-forward Networks
• Synthetic to Real Adaptation with Generative Correlation Alignment Networks

Cited by in corpus (7)

• Principled Weight Initialization for Hypernetworks
• Avatar-Net: Multi-scale Zero-shot Style Transfer by Feature Decoration
• A Style-Aware Content Loss for Real-time HD Style Transfer
• Regularization-Agnostic Compressed Sensing MRI Reconstruction with Hypernetworks
• EFANet: Exchangeable Feature Alignment Network for Arbitrary Style Transfer
• MVStylizer: An Efficient Edge-Assisted Video Photorealistic Style Transfer System for Mobile Phones
• Neural Comic Style Transfer: Case Study