EPFL-DrivingVQA/DrivingVQA

DrivingVQA是一个旨在帮助候选人准备法国驾驶理论考试的数据集，该考试包括理论测试和实践测试。数据集专注于视觉分析技能，包含基于图像的驾驶理论材料，由EPFL VITA和NLP实验室策划。数据集包含3931个样本，包括驾驶场景的图像、问题、可能的答案、正确答案、专家注释的解释以及相关实体的列表和边界框坐标。

DrivingVQA is a dataset designed to assist candidates preparing for the French driving theory exam, which includes both a theoretical and a practical test. The dataset focuses on visual analysis skills and contains image-based driving theory materials curated by EPFL VITA and NLP labs. It includes 3,931 samples with driving scene images, questions, possible answers, correct answers, expert-annotated explanations, and a list of relevant entities with bounding box coordinates.