HaVQA

HaVQA是首个为豪萨语设计的视觉问答（VQA）多模态数据集，由HausaNLP研究组创建。该数据集包含6022个手动翻译的英文问答对，关联1555张独特的图像，总计提供12044个黄金标准的英豪萨平行句，确保与相应视觉信息的语义匹配。HaVQA适用于多种NLP任务，如视觉问答、视觉问题引发、文本和多模态机器翻译，旨在丰富豪萨语的自然语言处理资源，支持在该语言中的VQA和多模态研究。

HaVQA is the first multimodal visual question answering (VQA) dataset developed for the Hausa language, created by the HausaNLP research group. This dataset includes 6,022 manually translated English question-answer pairs, linked to 1,555 unique images, and provides a total of 12,044 gold-standard English-Hausa parallel sentences, ensuring semantic alignment with the corresponding visual information. HaVQA supports a wide range of natural language processing (NLP) tasks, including visual question answering (VQA), visual question generation (VQG), text and multimodal machine translation. It aims to enrich the natural language processing resources for Hausa and support VQA and multimodal research in the Hausa language.