metadata

license: cc-by-4.0
pretty_name: 3DCompat200
size_categories:
  - 1K<n<10K

3DCoMPaT200 Dataset

The 3DCoMPaT200 dataset is a comprehensive collection of 3D objects with compositional part annotations. This repository contains various formats and versions of the dataset organized for different use cases.

📁 Directory Structure

2D Folder

Contains train, validation, and test data in tar format for 10 compositions:

Training set
Validation set
Test set

Each file contains 2D representations of the objects with their corresponding compositional part annotations.

HDF5 Folder

Contains point cloud data in HDF5 format with 2048 points per shape:

Single composition datasets (train/val/test)
10 composition datasets (train/val/test)

The HDF5 files are optimized for efficient loading and processing of point cloud data.

Challenge Folder

Contains grounding prompts used for the Grounded Segmentation Challenge. These prompts are designed to evaluate models' ability to perform semantic segmentation based on natural language descriptions.

Compat200.zip

Contains the original 3D object files in GLTF format:

Training set objects
Validation set objects Note: Test set objects are not included in this file.

🔍 Dataset Details

Number of points per shape: 2048
Number of compositions: 1 and 10 variants
File formats: TAR, HDF5, GLTF

🚀 Usage Instructions

For detailed instructions on how to use the dataset, including code examples and utility functions, please visit our GitHub repository: https://github.com/3DCoMPaT200/3DCoMPaT200

The repository contains loaders, rendering tools, and example code to help you get started with the dataset.

📝 Citation

If you use our dataset, please cite the three following references:

@inproceedings{ahmed2024dcompat,
title={3{DC}o{MP}aT200: Language Grounded Large-Scale 3D Vision Dataset for Compositional Recognition},
author={Mahmoud Ahmed and Xiang Li and Arpit Prajapati and Mohamed Elhoseiny},
booktitle={The Thirty-eight Conference on Neural Information Processing Systems Datasets and Benchmarks Track},
year={2024},
url={https://openreview.net/forum?id=L4yLhMjCOR}
}

@article{slim2023_3dcompatplus,
    title={3DCoMPaT++: An improved Large-scale 3D Vision Dataset
    for Compositional Recognition},
    author={Habib Slim, Xiang Li, Yuchen Li,
    Mahmoud Ahmed, Mohamed Ayman, Ujjwal Upadhyay
    Ahmed Abdelreheem, Arpit Prajapati,
    Suhail Pothigara, Peter Wonka, Mohamed Elhoseiny},
    year={2023}
}

@article{li2022_3dcompat,
    title={3D CoMPaT: Composition of Materials on Parts of 3D Things},
    author={Yuchen Li, Ujjwal Upadhyay, Habib Slim,
    Ahmed Abdelreheem, Arpit Prajapati,
    Suhail Pothigara, Peter Wonka, Mohamed Elhoseiny},
    journal = {ECCV},
    year={2022}
}