SimpsonsVQA
收藏资源简介:
SimpsonsVQA数据集由迪肯大学创建,专门用于视觉问答(VQA)任务,特别是基于《辛普森一家》电视节目的卡通图像。该数据集包含约23,000张图像和166,533对问答,旨在促进探究式学习,并解决现有模型在卡通图像上的表现不足问题。数据集通过自动化和人工审查相结合的方式创建,涵盖了从图像描述生成到问答对评估的完整流程。SimpsonsVQA的应用领域包括教育工具和辅助技术,旨在通过AI系统支持早期教育和认知障碍个体的学习。
The SimpsonsVQA dataset was developed by Deakin University, exclusively for visual question answering (VQA) tasks centered on cartoon images from the television series *The Simpsons*. This dataset comprises roughly 23,000 images and 166,533 question-answer pairs, with the goals of facilitating inquiry-based learning and addressing the subpar performance of current models when handling cartoon images. The dataset was constructed via a hybrid approach of automated procedures and human review, covering the full workflow ranging from image caption generation to question-answer pair evaluation. Application scenarios of SimpsonsVQA encompass educational tools and assistive technologies, with the objective of supporting early childhood education and learning for individuals with cognitive impairments through AI systems.




