Pluggable Style Representation Learning for Multi-Style Transfer
This addresses the problem of deploying style transfer on resource-limited devices by improving scalability and efficiency, though it is incremental in nature.
The paper tackles the challenge of scaling multi-style image transfer to diverse styles without increasing computational cost by decoupling style modeling and transferring, achieving state-of-the-art performance in accuracy and efficiency.
Due to the high diversity of image styles, the scalability to various styles plays a critical role in real-world applications. To accommodate a large amount of styles, previous multi-style transfer approaches rely on enlarging the model size while arbitrary-style transfer methods utilize heavy backbones. However, the additional computational cost introduced by more model parameters hinders these methods to be deployed on resource-limited devices. To address this challenge, in this paper, we develop a style transfer framework by decoupling the style modeling and transferring. Specifically, for style modeling, we propose a style representation learning scheme to encode the style information into a compact representation. Then, for style transferring, we develop a style-aware multi-style transfer network (SaMST) to adapt to diverse styles using pluggable style representations. In this way, our framework is able to accommodate diverse image styles in the learned style representations without introducing additional overhead during inference, thereby maintaining efficiency. Experiments show that our style representation can extract accurate style information. Moreover, qualitative and quantitative results demonstrate that our method achieves state-of-the-art performance in terms of both accuracy and efficiency. The codes are available in https://github.com/The-Learning-And-Vision-Atelier-LAVA/SaMST.