Release 4K360Vid dataset on Hugging Face

[NielsRogge]

Hi @L-li 🤗

I'm Niels and work as part of the open-source team at Hugging Face. I discovered your work through Hugging Face's daily papers as yours got featured: https://huggingface.co/papers/2603.04291.
The paper page lets people discuss about your paper and lets them find artifacts about it, and it's great to see that you've already hosted the CubeComposer checkpoints on the Hub! You can also claim the paper as yours which will show up on your public profile at HF, and add your GitHub and project page URLs.

Would you like to host the 4K360Vid dataset you've introduced on https://huggingface.co/datasets as well?
I noticed in your README that you plan to release the filtered version with face-wise captions later. Hosting on Hugging Face will give you more visibility/enable better discoverability, and will also allow people to do:

from datasets import load_dataset

dataset = load_dataset("TencentARC/4K360Vid")

If you're down, leaving a guide here: https://huggingface.co/docs/datasets/loading.
We also support Webdataset, which is particularly useful for large-scale video datasets: https://huggingface.co/docs/datasets/en/loading#webdataset.

Besides that, there's the dataset viewer which allows people to quickly explore the videos and captions directly in the browser.

After uploaded, we can also link the dataset to the paper page so people can discover your work.

Let me know if you're interested or need any guidance!

Kind regards,

Niels

[Red-Fairy]

Hi @NielsRogge.

I'm the author of the Argus paper. As noted in this CubeComposer paper, the paper used a subset of our proposed YouTube-sourced dataset. According to YouTube policy, it is prohibited to re-distribute the videos without premission.

If you find it useful, you can find the data used in our Argus paper in our repo with a detailed instruction on how to download and extract the clip we used. I believe the author of CubeComposer will also release the subset they used by providing the cilp_id and start/end time stamp.

Best,
Rundong

[lg-li]

Hi @NielsRogge, and thanks @Red-Fairy as well for the clarification!

Following this, I’ve now uploaded the dataset (with clip_ids, timestamps, and captions) to Hugging Face: https://huggingface.co/datasets/l-li/4K360Vid.

Thanks again for your help!

[NielsRogge]

Thanks so much for uploading the 4K360Vid dataset to Hugging Face and linking it both to the paper page and in your README. The dataset is now discoverable on the paper page at https://huggingface.co/papers/2603.04291, and users can easily find and use it.

If you haven't already, feel free to claim the paper by clicking on your name on the paper page. This will display the paper on your Hugging Face profile.

Thanks again for making your work openly accessible