MSR-VTT and UCF-101

The dataset used in the paper is MSR-VTT and UCF-101, two public datasets for video-text generation. MSR-VTT contains 4,900 videos with 20 manually annotated captions for each video, while UCF-101 contains 13,320 videos with 101 action categories.

Data and Resources

Cite this as

Kaifeng Gao, Jiaxin Shi, Hanwang Zhang, Chunping Wang, Jun Xiao (2024). Dataset: MSR-VTT and UCF-101. https://doi.org/10.57702/dvyul5ft

DOI retrieved: December 16, 2024

Additional Info

Field Value
Created December 16, 2024
Last update December 16, 2024
Author Kaifeng Gao
More Authors
Jiaxin Shi
Hanwang Zhang
Chunping Wang
Jun Xiao
Homepage https://github.com/Dawn-LX/Causal-VideoGen