# Dataset Preparation

We follow [UniFormerV2](https://github.com/OpenGVLab/UniFormerV2/) to prepare the datasets. All the data files can be found [here](https://drive.google.com/drive/folders/17VB-XdF3Kfr9ORmnGyXCxTMs86n0L4QL?usp=sharing), including:
- [Kinetics-400/600/700](https://www.deepmind.com/open-source/kinetics)
- [Kinetics-710](https://github.com/OpenGVLab/UniFormerV2/blob/main/DATASET.md)
- [Moments in Time V1](http://moments.csail.mit.edu/)
- [Something-Something V1&V2](https://developer.qualcomm.com/software/ai-datasets/something-something)
- [ANet](http://activity-net.org/)
- [HACS](http://hacs.csail.mit.edu/)

For K-Mash, we merge the above videos and other self-collected videos.