---
license: cc0-1.0
---

# [ECCV 2024] MinD-3D: Reconstruct High-quality 3D objects in Human Brain

[![ArXiv](https://img.shields.io/badge/ArXiv-2312.07485-b31b1b.svg?logo=arXiv)](https://arxiv.org/abs/2312.07485)

## Overview
MinD-3D aims to reconstruct high-quality 3D objects based on fMRI data. 

## Repository Structure
- **annotations**: Contains metadata and annotations related to the fMRI data for each subject.
- **sub-00xx**: Each folder corresponds to a specific subject and includes their respective raw and processed fMRI data.
- **stimuli.tar**: A tar archive of all videos shown to subjects during the fMRI scans. This file includes the stimuli used across different sessions and is critical for reproducibility of the study findings.
- **camera_pose.zip**: The camera pose for each frame in the videos (each containing 192 frames) in the stimuli.

## Data Description
- **raw_data**: Raw fMRI data collected directly from the imaging machine.
- **npy_data**: Processed data. We utilized fMRIPrep and the methodologies described in our paper to derive and store the data in NumPy format (.npy).


## Citation

If you find our paper useful for your research and applications, please cite using this BibTeX:

```
@misc{gao2023mind3d,
  title={MinD-3D: Reconstruct High-quality 3D objects in Human Brain}, 
  author={Jianxiong Gao and Yuqian Fu and Yun Wang and Xuelin Qian and Jianfeng Feng and Yanwei Fu},
  year={2023},
  eprint={2312.07485},
  archivePrefix={arXiv},
  primaryClass={cs.CV}
}
```