HiFiHuman Dataset

File Splitting and Local Reconstruction Guide

Due to the large size of individual .zip files (often larger than 1GB), all data in this dataset is uploaded to Hugging Face in split parts to ensure stable uploads, reliable downloads, and better compatibility with different network environments.

This document explains how the dataset is organized and how to reconstruct the original files locally after downloading.

---

Dataset Structure

Each original zip file is split into multiple binary parts before upload.
On Hugging Face, the directory structure looks like this:

TaoGS_Dataset/
├── 4K_Actor1_Magic_1/
│   ├── 4K_Actor1_Magic_1.zip.partaa
│   ├── 4K_Actor1_Magic_1.zip.partab
│   ├── 4K_Actor1_Magic_1.zip.partac
│   └── ...
├── 4K_Actor2_Change_Clothes_1/
│   ├── 4K_Actor2_Change_Clothes_1.zip.partaa
│   ├── 4K_Actor2_Change_Clothes_1.zip.partab
│   └── ...
└── ...

Each *.zip.part* file is a consecutive binary chunk generated using the Unix split command.
All parts must be merged in order to recover the original zip file.

---

Step 1: Download the Dataset Parts

You may download the required parts using the Hugging Face CLI or other supported tools.

Recommended: Hugging Face CLI

Install the CLI tool:

pip install -U huggingface_hub

Download a specific sequence:

hf download moqiyinlun1/HiFiHuman \
  --repo-type dataset \
  --local-dir ./HiFiHuman \
  --include "TaoGS_Dataset/4K_Actor1_Magic_1/*"

Downloading only the required subdirectory is strongly recommended to avoid unnecessary data transfer.

---

Step 2: Reconstruct the Original Zip File

Navigate to the directory containing the downloaded parts:

cd HiFiHuman/TaoGS_Dataset/4K_Actor1_Magic_1

Merge all parts into the original zip file:

cat 4K_Actor1_Magic_1.zip.parta* > 4K_Actor1_Magic_1.zip

The wildcard parta* ensures that the parts are concatenated in the correct order (partaa, partab, partac, ...).

---

Step 3: Verify the Reconstructed File (Optional)

To verify that the reconstructed file is valid:

unzip -t 4K_Actor1_Magic_1.zip

If no errors are reported, the reconstruction was successful.

---

Step 4: Extract the Dataset

Once the zip file is reconstructed:

unzip 4K_Actor1_Magic_1.zip

You can now use the dataset as intended.

---

Windows Users

On Windows (PowerShell), use the following command to merge the parts:

copy /b 4K_Actor1_Magic_1.zip.part* 4K_Actor1_Magic_1.zip

---

Notes

• Make sure all part files are fully downloaded before merging.
• Missing or corrupted parts will result in an invalid zip file.
• Do not attempt to unzip individual .part files.

---

Contact

If you encounter issues while downloading, reconstructing, or using the dataset, please open an issue on the Hugging Face repository or contact the shenzhh2025@shanghaitech.edu.cn directly.

---

Citation

If you find this dataset useful for your research, please cite these works using the following BibTeX entry.

@misc{jiang2025topology,
    title={Topology-Aware Optimization of Gaussian Primitives for Human-Centric Volumetric Videos}, 
    author={Yuheng Jiang and Chengcheng Guo and Yize Wu and Yu Hong and Shengkun Zhu and Zhehao Shen and Yingliang Zhang and Shaohui Jiao and Zhuo Su and Lan Xu and Marc Habermann and Christian Theobalt},
    year={2025},
    eprint={2509.07653},
    archivePrefix={arXiv},
    primaryClass={cs.GR},
    url={https://arxiv.org/abs/2509.07653}, 
}

@article{jiang2024robust,
    title={Robust dual gaussian splatting for immersive human-centric volumetric videos},
    author={Jiang, Yuheng and Shen, Zhehao and Hong, Yu and Guo, Chengcheng and Wu, Yize and Zhang, Yingliang and Yu, Jingyi and Xu, Lan},
    journal={ACM Transactions on Graphics (TOG)},
    volume={43},
    number={6},
    pages={1--15},
    year={2024},
    publisher={ACM New York, NY, USA}
}

@InProceedings{Jiang2025reperformer,
    author={Jiang, Yuheng and Shen, Zhehao and Guo, Chengcheng and Hong, Yu and Su, Zhuo and Zhang, Yingliang and Habermann, Marc and Xu, Lan},
    title={RePerformer: Immersive Human-centric Volumetric Videos from Playback to Photoreal Reperformance},
    booktitle={Proceedings of the Computer Vision and Pattern Recognition Conference (CVPR)},
    month={June},
    year={2025},
    pages={11349-11360}
}

@InProceedings{Jiang_2024_CVPR,
 author = {Jiang, Yuheng and Shen, Zhehao and Wang, Penghao and Su, Zhuo and Hong, Yu and Zhang, Yingliang and Yu, Jingyi and Xu, Lan}, 
 title = {HiFi4G: High-Fidelity Human Performance Rendering via Compact Gaussian Splatting}, 
 booktitle = {Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)}, 
 month = {June}, 
 year = {2024}, 
 pages = {19734-19745} 
 }