Webvid-10M

The dataset used for training the video model consists of Webvid-10M, a large-scale dataset of short videos with textual descriptions.

BibTex: