StyTr2: Image Style Transfer with Transformers

By Yingying Deng, Fan Tang, Weiming Dong, Chongyang Ma, Xingjia Pan, Lei Wang, Changsheng Xu. Computer Vision and Pattern Recognition, 2021

433 citations recorded by the source corpus.

The goal of image style transfer is to render an image with artistic features guided by a style reference while maintaining the original content. Owing to the locality in convolutional neural networks (CNNs), extracting and maintaining the global information of input images is difficult. Therefore, traditional neural style transfer methods face biased content representation. To address this critical issue, we take long-range dependencies of input images into account for image style transfer by proposing a transformer-based approach called StyTr2. In contrast with visual transformers for other vision tasks, StyTr2 contains two different transformer encoders to generate domain-specific sequences for content and style, respectively. Following the encoders, a multi-layer transformer decoder is adopted to stylize the content sequence according to the style sequence. We also analyze the deficiency of existing positional encoding methods and propose the content-aware positional encoding (CAPE), which is scale-invariant and more suitable for image style transfer tasks. Qualitative and quantitative experiments demonstrate the effectiveness of the proposed StyTr2 compared with state-of-the-art CNN-based and flow-based approaches. Code and models are available at https://github.com/diyiiyiii/StyTR-2.

Subfields: Computer Vision and Pattern Recognition

View Paper: https://www.semanticscholar.org/paper/1079b5459291ec73fc0a2d0692d38eaa97cb41a6

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computer Vision and Pattern Recognition

1. Fan Tang
2. Changsheng Xu
3. Sangeek Hyun
4. Chongyang Ma
5. Yu-xin Zhang
6. Nisha Huang
7. Jiwoo Chung
8. Jae-Pil Heo

Citations in the corpus

• Inversion-based Style Transfer with Diffusion Models (2022, methodology): “[6] propose a transformer-based method (StyTr(2)) to avoid the biased content representation in style transfer by taking long-range dependencies of input images into account.”
• Style Injection in Diffusion: A Training-Free Approach for Adapting Large-Scale Diffusion Models for Style Transfer (2023, positive): “Differently, InST synthesizes the realistic images, while struggling to transfer style ( 1 st row) or change content of image ( 2 nd , 3 rd rows).”