Thanks to visit codestin.com
Credit goes to github.com

Skip to content

zcai0612/UP2You

Repository files navigation

TL;DR: Tuning-free, Fast Reconstruct Yourself from Unconstrained Photo Collections

UP2You Teaser

Getting Started

Installation

  1. Clone UP2You.
git clone https://github.com/zcai0612/UP2You.git
cd UP2You
  1. Create the environment.
conda create -n up2you python=3.10
conda activate up2you

# torch
pip install torch==2.4.1 torchvision==0.19.1 torchaudio==2.4.1 --index-url https://download.pytorch.org/whl/cu118

# kaolin
pip install kaolin==0.17.0 -f https://nvidia-kaolin.s3.us-east-2.amazonaws.com/torch-2.4.1_cu118.html

pip install -r requirements.txt

# https://github.com/cnr-isti-vclab/meshlab/issues/1461
conda install -y libffi==3.3

# install pytorch3d, download from conda
wget https://anaconda.org/pytorch3d/pytorch3d/0.7.8/download/linux-64/pytorch3d-0.7.8-py310_cu118_pyt241.tar.bz2
conda install pytorch3d-0.7.8-py310_cu118_pyt241.tar.bz2

Download

Download pretrained_models and human_models from huggingface Co2y/UP2You, and put them into the project directory. You can refer to following commands:

export HF_ENDPOINT="https://hf-mirror.com" # Optional

hf download Co2y/UP2You --local-dir ./src # download models

mv ./src/human_models ./
mv ./src/pretrained_models ./

rm -rf ./src

Inference

To run the inference pipeline, you can use the following command:

python inference_low_gpu.py \
    --base_model_path stabilityai/stable-diffusion-2-1-base  \
    --segment_model_name ZhengPeng7/BiRefNet \
    --data_dir examples \
    --output_dir outputs \

or you can just use run.sh:

bash run.sh 

Here we provide an example, where examples is the folder of unconstrained photos and outputs is the output directory of generated results.

Acknowledgements

Our code is based on the following awesome repositories and datasets:

We thank the authors for releasing their code and data !

We thank Siyuan Yu for the help in Houdini Simulation, Shunsuke Saito, Dianbing Xi, Yifei Zeng for the fruitful discussions, and the members of Endless AI Lab for their help on data capture and discussions.

Citation

If you find our work useful, please cite:

@article{cai2025up2you,
  title={UP2You: Fast Reconstruction of Yourself from Unconstrained Photo Collections},
  author={Cai, Zeyu and Li, Ziyang and Li, Xiaoben and Li, Boqian and Wang, Zeyu and Zhang, Zhenyu and Xiu, Yuliang},
  journal={arXiv preprint arXiv:2509.24817},
  year={2025}
}

About

No description, website, or topics provided.

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages