TGIF-QA
收藏arXiv2025-09-30 收录
下载链接：
https://github.com/yunseokjang/tgif-qa
下载链接
链接失效反馈官方服务：
资源简介：
该数据集名为TGIF-QA，包含了一系列多样化的问答任务，如统计动作的重复次数、关于动作的多选题、强调视频中的时间过渡以及关于视频帧视觉内容的开放式问题。对于TGIF-QA数据集，特征提取是使用预训练的ResNet模型完成的。该任务的类型是视频问答。
The dataset is named TGIF-QA, which includes a diverse set of question answering (QA) tasks, such as counting the number of action repetitions, multiple-choice questions about actions, tasks emphasizing temporal transitions in videos, and open-ended questions regarding the visual content of video frames. For the TGIF-QA dataset, feature extraction is performed using a pre-trained ResNet model. The task type of this dataset is video question answering.
提供机构：
TGIF-QA
