Reproduce “SemRAG: Semantic-Guided Sentence-Level Retrieval-Augmented Sign Language Production” 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

Xiao Liu, Shiwei Gan, Yafeng Yin, Peishuo Liu, Bowen Guo, Kuizhuang Liu, Cong Wang, Lei Xie

SemRAG generates sign-pose sequences from text by retrieving a sentence-level signing example, using sign-language translation attention to filter relevant temporal regions, and gating the reference into the generator. On Phoenix14T test, its b variant reaches 12.29 BLEU-4 and 22.47 MPJPE, compared with SOKE’s 10.43 and 22.94. On CSL-Daily, BLEU-4 is higher, but MPJPE is slightly worse: 22.75 versus 22.50.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.