Artistic Neural Style Transfer Algorithms with Activation Smoothing
This work addresses the enhancement of image stylization quality for applications in digital art and media, but it is incremental as it builds on existing NST techniques.
The paper tackled the problem of improving artistic neural style transfer by implementing various NST methods and exploring ResNet with activation smoothing, resulting in a significant quality improvement in stylization outcomes.
The works of Gatys et al. demonstrated the capability of Convolutional Neural Networks (CNNs) in creating artistic style images. This process of transferring content images in different styles is called Neural Style Transfer (NST). In this paper, we re-implement image-based NST, fast NST, and arbitrary NST. We also explore to utilize ResNet with activation smoothing in NST. Extensive experimental results demonstrate that smoothing transformation can greatly improve the quality of stylization results.