README.md

Objectron Videos Mirror

This repository is a videos-only mirror of the official Objectron dataset, prepared for hosting on Hugging Face.

Official source repository:

• https://github.com/google-research-datasets/Objectron

Purpose

• Provide a clean and upload-friendly copy of Objectron video files.
• Keep directory layout aligned with official dataset conventions.
• Simplify distribution for downstream training and research workflows.

What Is Included

This mirror currently contains only video files.

Included:

• videos/<class>/batch-<i>/<j>/<video>.MOV

Not included in this mirror:

• annotation protobufs (for example geometry.pbdata)
• AR metadata protobufs
• tf.records / sequence examples
• index files and train/test split files
• parsing/evaluation scripts

For full dataset assets and tooling, use the official repository and storage paths.

Directory Layout

The video files follow the official Objectron layout pattern:

• videos/class/batch-i/j/video.MOV

Current class folders may include:

• bike
• book
• bottle
• camera
• cereal_box
• chair
• cup
• laptop
• shoe

License

This repository follows the official Objectron licensing terms.

Objectron is released under:

• Computational Use of Data Agreement 1.0 (C-UDA-1.0)
• https://github.com/microsoft/Computational-Use-of-Data-Agreement

A copy of the license is included in LICENSE.

Attribution

If you use Objectron data, please cite the official Objectron paper and follow attribution guidance from the official repository:

• https://github.com/google-research-datasets/Objectron

Acknowledgment

We thank the Objectron team and the official maintainers for providing this dataset and related resources. These contributions were instrumental in the successful completion of our work: ConsID-Gen.

Objectron is a large-scale, object-centric video dataset with pose annotations and has made important contributions to 3D understanding and related vision research.

This repository is only a videos-only mirror for easier access and distribution.

Disclaimer

• This repository is not an official Google release.
• We cannot guarantee that the number of videos in this mirror exactly matches the counts reported in the original Objectron paper or official storage.
• The contents here only include video files available from our local download process.

Citation

If you found the original Objectron dataset useful, please cite the official paper.

@article{objectron2021,
    title={Objectron: A Large Scale Dataset of Object-Centric Videos in the Wild with Pose Annotations},
    author={Adel Ahmadyan, Liangkai Zhang, Artsiom Ablavatski, Jianing Wei, Matthias Grundmann},
    journal={Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition},
    year={2021}
}

This is not an officially supported Google product. If you have any question, you can email us at objectron@google.com or join our mailing list at objectron@googlegroups.com.

@misc{wu2026considgenviewconsistentidentitypreservingimagetovideo,
    title={ConsID-Gen: View-Consistent and Identity-Preserving Image-to-Video Generation}, 
    author={Mingyang Wu and Ashirbad Mishra and Soumik Dey and Shuo Xing and Naveen Ravipati and Hansi Wu and Binbin Li and Zhengzhong Tu},
    year={2026},
    eprint={2602.10113},
    archivePrefix={arXiv},
    primaryClass={cs.CV},
    url={https://arxiv.org/abs/2602.10113},
}