Are the video titles and tags provided by the creators, or generated by some video-language models?

Question

Are the video titles and tags provided by the creators, or generated by some video-language models?

Closed this issue 2 months ago · 1 comments

I am curious about how these features, such as titles and tags, are obtained. Are they manually provided by the video creators, or are they automatically generated through advanced video-text models?

Answer 1 · 2024-10-17T13:02:35.000Z

Hi @CleverPaul , all video and related data are created by the human creators, no generation data involved in the dataset. Please see our paper for the process of dataset establishment.