Retrieval-Augmented Long-Context Translation for Cultural Image Captioning: Gators submission for AmericasNLP 2026 shared task
The University of Florida Gators submitted a paper for the AmericasNLP 2026 shared task focusing on cultural image captioning for Indigenous languages. Their approach involves a two-stage pipeline that significantly improves captioning performance for languages like Bribri, Guaraní, and Orizaba Nahuatl. The submission achieved the overall winner status in the shared task and performed well in human evaluations.
- ▪The submission utilizes a two-stage pipeline generating an intermediate Spanish caption before producing the target-language caption.
- ▪Improvements over the shared task baseline were noted, with gains of 164.1%, 131.7%, and 122.6% for Bribri, Guaraní, and Orizaba Nahuatl, respectively.
- ▪The Gators' submission was the overall winner of the shared task and placed second in human evaluations.
Opening excerpt (first ~120 words) tap to expand
Computer Science > Computation and Language arXiv:2605.20626 (cs) [Submitted on 20 May 2026] Title:Retrieval-Augmented Long-Context Translation for Cultural Image Captioning: Gators submission for AmericasNLP 2026 shared task Authors:Aashish Dhawan, Christopher Driggers-Ellis, Dzmitry Kasinets, Daisy Zhe Wang, Christan Grant View a PDF of the paper titled Retrieval-Augmented Long-Context Translation for Cultural Image Captioning: Gators submission for AmericasNLP 2026 shared task, by Aashish Dhawan and 4 other authors View PDF HTML (experimental) Abstract:We present the University of Florida Gators submission to the AmericasNLP 2026 shared task on cultural image captioning for Indigenous languages.
…
Excerpt limited to ~120 words for fair-use compliance. The full article is at arXiv cs.AI.
Discussion
0 commentsMore from arXiv cs.AI
