VideoDatasets/SpatialVID-RAW

SpatialVID: A Large-Scale Video Dataset with Spatial Annotations

Jiahao Wang^1*  Yufeng Yuan^1*  Rujie Zheng^1*  Youtian Lin¹  Jian Gao¹  Lin-Zhuo Chen¹ 

Yajie Bao¹  Yi Zhang¹  Chang Zeng¹  Yanxi Zhou¹  Xiaoxiao Long¹  Hao Zhu¹ 

Zhaoxiang Zhang²  Xun Cao¹  Yao Yao^1†

¹Nanjing University  ²Institute of Automation, Chinese Academy of Science 

*Equal Contribution  †Corresponding Author

CVPR 2026

       

SpatialVID-RAW

Thank you for your interest in our work. We have received numerous requests from the research community for access to the raw data. To support further research, we are providing the YouTube video IDs and their corresponding timestamps used in our dataset.

Due to potential copyright considerations, this metadata is available through a gated dataset on Hugging Face rather than being fully public. To gain access, you will be required to submit a request. This process helps ensure the responsible and appropriate use of the dataset. We appreciate your understanding and cooperation.

By downloading and using this dataset, you agree to comply with the terms of the LICENSE AGREEMENT.

Dataset Description

Note that group id 0 means the clip is filtered out in our previous processing. So these clips are not included in the SpatialVID-HQ or SpatialVID dataset.

The provided metadata is categorized as follows:

• Short Clips: This set contains timestamps for video segments that are between 3 and 15 seconds long. A portion of these short clips, determined by their "group id," were used for the annotations reported in our paper. Clips with a "group id" of 0 were filtered out in our previous processing steps.

  Field Name	Description
  id	Unique identifier for each video clip (matches the annotation folder name).
  group id	Identifier of the group the video clip belongs to (e.g., group_0001).
  YouTube id	Unique identifier for the YouTube video (e.g., dQw4w9WgXcQ).
  timestamp_start	Start time of the video segment (in microseconds).
  timestamp_end	End time of the video segment (in microseconds).
  fps	Frames per second (FPS) of the video clip.
  aesthetic score	Subjective score evaluating the video’s visual aesthetics (0–10).
  luminance score	Score measuring the video’s overall brightness.
  motion score	Score quantifying the intensity of motion in the video (0–20).

• Long Clips: To accommodate research that requires extended video context, we also provide timestamps without the 15-second length constraint. Please note that these longer versions do not have separate annotations. However, you can either use our provided annotation process to create your own or utilize the caption information from the corresponding short clips.

  Field Name	Description
  YouTube id	Unique identifier for the YouTube video (e.g., dQw4w9WgXcQ).
  timestamp_start	Start time of the video segment (in microseconds).
  timestamp_end	End time of the video segment (in microseconds).
  fps	Frames per second (FPS) of the video clip.
  duration	Duration of the video segment (in microseconds).

Download Tool

To facilitate the download of video clips based on the provided metadata, we have developed a Python script that allows you to easily download the specified video segments. Please refer to the download_YouTube.py in our GitHub repo.

License

SpatialVID is released under the Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License (CC-BY-NC-SA 4.0). By using this dataset, you agree to the following terms:

1. Attribution: You must credit the original source of the dataset.
2. Non-Commercial Use: The dataset may not be used for commercial purposes.
3. ShareAlike: Any modified or derived works must be released under the same license.

For the full license text, visit: https://creativecommons.org/licenses/by-nc-sa/4.0/legalcode.

If you are the copyright owner of any video in our dataset and you need it to be removed, please contact us, and we will remove the video samples from our dataset / Github / project webpage / technical presentation as soon as possible.

Citation

If you use SpatialVID in your research, please cite our work using the following format (citation will be updated upon paper publication):

@article{wang2025spatialvid,
  title={Spatialvid: A large-scale video dataset with spatial annotations},
  author={Wang, Jiahao and Yuan, Yufeng and Zheng, Rujie and Lin, Youtian and Gao, Jian and Chen, Lin-Zhuo and Bao, Yajie and Zhang, Yi and Zeng, Chang and Zhou, Yanxi and others},
  journal={arXiv preprint arXiv:2509.09676},
  year={2025}
}