Reproduce “Self-Prompting Diffusion Transformer for Open-Vocabulary Scene Text Editing via In-Context Learning” 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
Hongxi Li, Tong Wang, Chengjing Wu, Tianbao Liu, Jiangtao Yao, Xiaochao Qu, Xinxiao Wu, Luoqi Liu and 1 more
The authors introduce a scene-text editing method that changes words while aiming to preserve their original appearance. It conditions a diffusion transformer on a rendered template of the requested letters and a crop of the source text, then refines training on paired edits. On English AnyText-benchmark data, it reaches 0.8857 sentence accuracy versus TextFlux’s 0.8231; this comparison is English-only.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.