README.md

metadata

language:
  - en
license: cc-by-nc-sa-4.0
pipeline_tag: text-to-3D
tags:
  - text-to-3D
  - dataset
  - annotation
  - captioning
extra_gated_heading: Acknowledge license to accept the repository
extra_gated_prompt: "### MARVEL-40M+ Community License Agreement\n\n [![License: CC BY-NC-SA 4.0](https://img.shields.io/badge/License-CC%20BY--NC--SA%204.0-lightgrey.svg)](https://creativecommons.org/licenses/by-nc-sa/4.0/)\n \n This named Dataset, **MARVEL-40M+**, was developed by [Deutsche Forschungszentrum für Kuntlische Inlligenz (DFKI GmbH)] (Germany). It is licensed under the Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License.\n \n ## Attribution-NonCommercial-ShareAlike 4.0 International (CC BY-NC-SA 4.0)\n You are free to:\n - **Share**: copy and redistribute the material in any medium or format.\n - **Adapt**: remix, transform, and build upon the material.\n \n\n Under the following terms:\n - **Attribution**: You must give appropriate credit, provide a link to the license, and indicate if changes were made. You may do so in any reasonable manner, but not in any way that suggests the licensor endorses you or your use.\n - **NonCommercial**: You may not use the material for commercial purposes.\n - **ShareAlike**: If you remix, transform, or build upon the material, you must distribute your contributions under the same license as the original.\n \n \n \n To view a copy of this license, visit [https://creativecommons.org/licenses/by-nc-sa/4.0/](https://creativecommons.org/licenses/by-nc-sa/4.0/).\n\n For inquiries or additional information, contact:\n \n \n \n - [DFKI GmbH\_ Contact Information]\n   \n   Deutsches Forschungszentrum Fur Kunstliche Intelligenz GmbH\n   Trippstadter Str. 122\n   67663 Kaiserslautern\n   Germany\n   Phone: +49 631 20575 0\n   link: https://www.dfki.de/web\n"
extra_gated_fields:
  First Name: text
  Last Name: text
  Country: country
  Affiliation: text
  I want to use this dataset for:
    type: select
    options:
      - Research
      - Education
      - label: Other
        value: other
  I agree to use this dataset for non-commercial use ONLY: checkbox

MARVEL-40M+
Multi-Level Visual Elaboration for
High-Fidelity Text-to-3D Content Creation

🧑‍💻 Sankalp Sinha   ·   🧑‍💻 Mohammad Sadil Khan   ·   Muhammad Usama   ·   Shino Sam   ·   Didier Stricker   ·   Sk Aziz Ali   ·   Muhammad Zeshan Afzal

🧑‍💻Equally Contributing First Authors

CVPR 2025

✅ Tasks

• Objaverse-XL Release.

• GSO2, ABO, Toys4k, Pix3D, OmniObject3D Release.

• Shapenet Release.

• Objaverse 1.0 Release.

🗂️ Folder Description

Annotation (Objaverse 1.0)

• id: Model identifier.
• filtered_name: Filtered Name (Using Mistral-Nemo).
• filtered_tags: Filtered Tags (Using Mistral-Nemo).
• filtered_description: Filtered Description (Using Mistral-Nemo).
• marvel_dense_description: Dense Multi-View Description (Generated by InternVL2-40M+).
• marvel_level_1: Comprehensive Annotation.
• marvel_level_2: Moderately Descriptive.
• marvel_level_3: Functional Semantic.
• marvel_level_4: Summary.
• marvel_level_5: Concise Tags.

Annotation (ShapeNet)

• id: Model identifier.
• synsetId: ShapeNet Synset Identifier.
• filtered_name: Filtered Name (Using Mistral-Nemo).
• filtered_tags: Filtered Tags (Using Mistral-Nemo).
• filtered_description: Filtered Description (Using Mistral-Nemo).
• marvel_dense_description: Dense Multi-View Description (Generated by InternVL2-40M+).
• marvel_level_1: Comprehensive Annotation.
• marvel_level_2: Moderately Descriptive.
• marvel_level_3: Functional Semantic.
• marvel_level_4: Summary.
• marvel_level_5: Concise Tags.

Annotation (GSO2, ABO, Toys4k, Pix3D, OmniObject3D)

• id: Model identifier.
• marvel_dense_description: Dense Multi-View Description (Generated by InternVL2-40M+).
• marvel_level_1: Comprehensive Annotation.
• marvel_level_2: Moderately Descriptive.
• marvel_level_3: Functional Semantic.
• marvel_level_4: Summary.
• marvel_level_5: Concise Tags.

📜 Citation

If you use this dataset in your work, please cite the following publication.

@inproceedings{sinha2025marvel,
  title={MARVEL-40M+: Multi-Level Visual Elaboration for High-Fidelity Text-to-3D Content Creation},
  author={Sinha, Sankalp and Khan, Mohammad Sadil and Usama, Muhammad and Sam, Shino and Stricker, Didier and Ali, Sk Aziz and Afzal, Muhammad Zeshan},
  booktitle={Proceedings of the Computer Vision and Pattern Recognition Conference},
  pages={8105--8116},
  year={2025}
}