zzlynxSG/SEA-Omni-Image

SEA-Omni Image Benchmark Southeast-Asia-oriented image benchmark released with our EMNLP 2026 paper (see Citation). It captures culturally grounded multimodal cues across English, Chinese, Malay, Indonesian and Thai; tasks cover captioning, summarisation, visual QA, OCR and multiple-choice QA. 10,005 test instances across 5 tasks, 5 languages (en, zh, id, ms, th). Images are sourced from the crowdsourced subset of SEA-VL (CC BY-SA 4.0) and synthetic OCR overlays. This benchmark… See the full description on the dataset page: https://huggingface.co/datasets/zzlynxSG/SEA-Omni-Image.

<iframe src="https://silytics.si/embed/item/61ae3fca-4165-4e83-8f0d-913b17209a53" width="420" height="190" style="border:0" loading="lazy" title="Silytics entry"></iframe>

Silytics shows public metadata from Hugging Face and links to the original. Information may be incomplete or out of date. Last refreshed 25 October 2026. Licenses and terms of the original apply to the work itself. Not financial, legal or professional advice.