junchen14 / 3DCoMPaT

Official repository for the 3DCoMPaT dataset (ECCV2022 Oral)

3D CoMPaT: Composition of Materials on Parts of 3D Things (ECCV 2022, oral)

Created by: Yuchen Li, Ujjwal Upadhyay, Habib Slim, Ahmed Abdelreheem, Arpit Prajapati, Suhail Pothigara, Peter Wonka, Mohamed Elhoseiny

Introduction

This is the official repository for 3DCoMPaT, "Composition of Materials on Parts of 3D Things", accepted at ECCV-2022 for an Oral presentation.

Dataset

To get the most out of this repository, please download the 3DCoMPaT dataset by filling this form.

Browser

You can browse the 3D models using the following link: 3D CoMPaT Browser

Compositional Benchmarks

• BPNet

• PointGroup

Non-Compositional Benchmarks

• 3D Shape and Part Classification

• 3D Sim2Real Transfer

• 2D/3D Material Segmentation

• 2D Material Tagging/ Shape Classification

Code release progress

• Add code and pretrained models for 3D Shape and Part Classification tasks
• Add code and pretrained models for Sim2Real Transfer tasks
• Add the 2D data-API
• Add code and pre-trained models for 2D Shape Classification
• Add code and pretrained models for BPNet
• Add code for 2D/3D Material Segmentation tasks
• Add the 3D data-API
• Add the evaluation code
• Add code and pre-trained models for 2D Material Tagging
• Provide examples for evaluating GCR task.
• Add pretrained models for 2D/3D Material Segmentation tasks
• Add code and pretrained models for PointGroup

Citation

If you find this work useful in your research, please consider citing:

@article{li20223dcompat,
    title={3D CoMPaT: Composition of Materials on Parts of 3D Things (ECCV 2022)},
    author={Yuchen Li, Ujjwal Upadhyay, Habib Slim,
    Ahmed Abdelreheem, Arpit Prajapati,
    Suhail Pothigara, Peter Wonka, Mohamed Elhoseiny},
    journal = {ECCV},
    year={2022}
}