YouTube thumbnails dataset
收藏arXiv2025-09-30 收录
下载链接:
https://github.com/mlpc-ucsd/BLIVA
下载链接
链接失效反馈官方服务:
资源简介:
该数据集包含了来自11个不同类别的YouTube缩略图与配套的问答集,旨在专门评估BLIVA模型在处理富含文本的视觉问题上的性能。该数据集适用于视觉问答(Vqa)任务。
This dataset contains YouTube thumbnails and their corresponding question-answer pairs across 11 distinct categories. It is specifically designed to evaluate the BLIVA model's performance on text-rich visual question answering tasks. This dataset is tailored for the visual question answering (VQA) task.



