StackOverflowVQA

StackOverflowVQA是由伊朗科技大学创建的视觉问答数据集，专注于软件相关的编程问题。该数据集包含451,394个问题，每个问题至少配有一张图片和多个全句答案，总计799,270个答案。数据集通过筛选和处理StackOverflow上的问题和答案创建，特别适用于训练视觉感知计算机助手。数据集中的图像多为计算机截图，问题和答案涉及编程知识，旨在通过图像和文本的结合提高问答系统的准确性。StackOverflowVQA is a visual question answering (VQA) dataset developed by Iran University of Science and Technology, focusing on software-related programming problems. This dataset contains 451,394 questions, each paired with at least one image and multiple full-sentence answers, with a total of 799,270 answers across the entire collection. Constructed by filtering and curating questions and answers from the StackOverflow platform, the dataset is specifically designed for training visually-aware computer assistants. The majority of images in the dataset are computer screenshots, and both the questions and answers cover programming-related knowledge. Its core objective is to improve the accuracy of question answering systems by combining visual and textual information.