Namronaldo2004/ViInfographicsVQA

ViInfographicsVQA是一个越南语视觉问答（VQA）数据集，由来自26个不同新闻平台的 infographic 构成。该数据集旨在支持多模态学习研究，提供基于现实世界视觉数据的各种问题和答案。数据集创建过程中严格遵循一定的规则和约束，以确保问题的质量和可答性。每个图像生成大约5个问题，包括3个基于文本的问题和2个非文本问题。问题和答案都有限长度的要求，并且答案需要包含直接的解释。

ViInfographicsVQA is a Vietnamese Visual Question Answering (VQA) dataset constructed from infographics sourced from 26 different news platforms. The dataset is designed to support research in multimodal learning by providing diverse questions and answers based on real-world visual data. The dataset creation process follows strict rules and constraints to ensure the quality and answerability of the questions. Approximately 5 questions are generated per image, including 3 text-based questions and 2 non-text questions. Both questions and answers have length constraints, and answers must include a direct explanation.