Reproduce “SAFIN: Arbitrary Style Transfer With Self-Attentive Factorized Instance Normalization” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Aaditya Singh, Shreeshail Hingane, Xinyu Gong, Zhangyang Wang
SAFIN uses self-attention to set spatially varying normalization parameters for transferring artistic style textures to relevant regions of a content image. It applies attention to normalized content and style features to infer regional scale and shift values, rather than directly transferring attention features. In examples it shows no obvious spillover compared with WCT and AdaIN; this is an illustrative comparison.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.