Skip to content
View stylegan-human's full-sized avatar

Block or report stylegan-human

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Please don't include any personal information such as legal names or email addresses. Maximum 100 characters, markdown supported. This note will be visible to only you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
stylegan-human/README.md

StyleGAN-Human: A Data-Centric Odyssey of Human Generation

Abstract: Unconditional human image generation is an important task in vision and graphics, which enables various applications in the creative industry. Existing studies in this field mainly focus on "network engineering" such as designing new components and objective functions. This work takes a data-centric perspective and investigates multiple critical aspects in "data engineering", which we believe would complement the current practice. To facilitate a comprehensive study, we collect and annotate a large-scale human image dataset with over 230K samples capturing diverse poses and textures. Equipped with this large dataset, we rigorously investigate three essential factors in data engineering for StyleGAN-based human generation, namely data size, data distribution, and data alignment. Extensive experiments reveal several valuable observations w.r.t. these aspects: 1) Large-scale data, more than 40K images, are needed to train a high-fidelity unconditional human generation model with vanilla StyleGAN. 2) A balanced training set helps improve the generation quality with rare face poses compared to the long-tailed counterpart, whereas simply balancing the clothing texture distribution does not effectively bring an improvement. 3) Human GAN models with body centers for alignment outperform models trained using face centers or pelvis points as alignment anchors. In addition, a model zoo and human editing applications are demonstrated to facilitate future research in the community.
Keyword: Human Image Generation, Data-Centric, StyleGAN

Jianglin Fu, Shikai Li, Yuming Jiang, Kwan-Yee Lin, Chen Qian, Chen Change Loy, Wayne Wu, and Ziwei Liu
[Demo Video] | [Project Page] | [Paper]

Updates

  • [07/04/2024] 🌟🌟🌟 Check out our new work in human foundation model -- CosmicMan at CVPR 2024!
  • [13/02/2023] 🌟🌟🌟 Now the human parsing and keypoints for SHHQ-1.0 are available!
  • [14/12/2022] 🔥🔥🔥We have released 3DHumanGAN, which is towards photo-realistic 3D-aware human generation!🔥🔥🔥
  • [28/09/2022] 🔥🔥🔥We have released a high-quality 3D human generative model EVA3D!🔥🔥🔥
  • [20/07/2022] SHHQ-1.0 dataset with 40K images is released! ✨
  • [15/06/2022] Data alignment and real-image inversion scripts are released.
  • [26/04/2022] Technical report released!
  • [22/04/2022] Technical report will be released before May.
  • [21/04/2022] The codebase and project page are created.

Data Download

The first version SHHQ-1.0, with 40K images is released. To download and use the dataset set, please read the instructions in Dataset.md

(We are currently facing large incoming applications, and we need to carefully verify all the applicants, please be patient, and we will reply to you as soon as possible.)

Model Zoo

Structure 1024x512 Metric Scores 512x256 Metric Scores
StyleGAN1 stylegan_human_v1_1024.pkl fid50k 3.79 to be released - -
StyleGAN2 stylegan_human_v2_1024.pkl fid50k_full 1.57 stylegan_human_v2_512.pkl fid50k_full 1.97
StyleGAN3 to be released - - stylegan_human_v3_512.pkl fid50k_full 2.54

Web Demo

Integrated into Huggingface Spaces 🤗 using Gradio. Try out the Web Demo for generation: Hugging Face Spaces and interpolation Hugging Face Spaces

We prepare a Colab demo to allow you to synthesize images with the provided models, as well as visualize the performance of style-mixing, interpolation, and attributes editing. The notebook will guide you to install the necessary environment and download pretrained models. The output images can be found in ./StyleGAN-Human/outputs/. Hope you enjoy!

Usage

System requirements

Installation

To work with this project on your own machine, you need to install the environmnet as follows:

conda env create -f environment.yml
conda activate stylehuman
# [Optional: tensorflow 1.x is required for StyleGAN1. ]
pip install nvidia-pyindex
pip install nvidia-tensorflow[horovod]
pip install nvidia-tensorboard==1.15

Extra notes:

  1. In case having some conflicts when calling CUDA version, please try to empty the LD_LIBRARY_PATH. For example:
LD_LIBRARY_PATH=; python generate.py --outdir=out/stylegan_human_v2_1024 --trunc=1 --seeds=1,3,5,7 
--network=pretrained_models/stylegan_human_v2_1024.pkl --version 2
  1. We found the following troubleshooting links might be helpful: 1., 2.

Train

The training scripts are based on the original stylegan1, stylegan2-ada, and stylegan3 with minor changes. Here we only provide the scripts with modifications for SG2 and SG3. You can replace the old files with the provided scripts to train. (assume SHHQ-1.0 is placed under data/)

Train Stylegan2-ada-pytorch with SHHQ-1.0

python train.py --outdir=training_results/sg2/ --data=data/SHHQ-1.0/ \
    --gpus=8 --aug=noaug --mirror=1 --snap=250 --cfg=shhq --square=False

Train Stylegan3 with SHHQ-1.0

python train.py --outdir=training_results/sg3/ --cfg=stylegan3-r --gpus=8 --batch=32 --gamma=12.4 \
    --mirror=1 --aug=noaug --data=data/SHHQ-1.0/ --square=False --snap=250

Pretrained models

Please put the downloaded pretrained models from above link under the folder 'pretrained_models'.

Generate full-body human images using our pretrained model

# Generate human full-body images without truncation
python generate.py --outdir=outputs/generate/stylegan_human_v2_1024 --trunc=1 --seeds=1,3,5,7 --network=pretrained_models/stylegan_human_v2_1024.pkl --version 2

# Generate human full-body images with truncation 
python generate.py --outdir=outputs/generate/stylegan_human_v2_1024 --trunc=0.8 --seeds=0-10 --network=pretrained_models/stylegan_human_v2_1024.pkl --version 2

# Generate human full-body images using stylegan V1
python generate.py --outdir=outputs/generate/stylegan_human_v1_1024 --network=pretrained_models/stylegan_human_v1_1024.pkl --version 1 --seeds=1,3,5

# Generate human full-body images using stylegan V3
python generate.py --outdir=outputs/generate/stylegan_human_v3_512 --network=pretrained_models/stylegan_human_v3_512.pkl --version 3 --seeds=1,3,5

Note: The following demos are generated based on models related to StyleGAN V2 (stylegan_human_v2_512.pkl and stylegan_human_v2_1024.pkl). If you want to see results for V1 or V3, you need to change the loading method of the corresponding models.

Interpolation

python interpolation.py --network=pretrained_models/stylegan_human_v2_1024.pkl  --seeds=85,100 --outdir=outputs/inter_gifs

Style-mixing image using stylegan2

python style_mixing.py --network=pretrained_models/stylegan_human_v2_1024.pkl --rows=85,100,75,458,1500 \\
    --cols=55,821,1789,293 --styles=0-3 --outdir=outputs/stylemixing 

Style-mixing video using stylegan2

python stylemixing_video.py --network=pretrained_models/stylegan_human_v2_1024.pkl --row-seed=3859 \\
    --col-seeds=3098,31759,3791 --col-styles=8-12 --trunc=0.8 --outdir=outputs/stylemixing_video

Aligned raw images

For alignment, we use openpose-pytorch for body-keypoints detection and PaddlePaddle for human segmentation. Before running the alignment script, few models need to be installed:

  1. download body_pose_model.pth and place it into openpose/model/.
  2. download and extract deeplabv3p_resnet50_os8_humanseg_512x512_100k_with_softmax into PP_HumanSeg/export_model/deeplabv3p_resnet50_os8_humanseg_512x512_100k_with_softmax.
  3. download and extract deeplabv3p_resnet50_os8_humanseg_512x512_100k into PP_HumanSeg/pretrained_model/deeplabv3p_resnet50_os8_humanseg_512x512_100k.
  4. install paddlepaddel: pip install paddleseg

Then you can start alignment:

python alignment.py --image-folder img/test/ --output-folder aligned_image/

Invert real image with PTI

Before inversion, please download our PTI weights: e4e_w+.pt into /pti/.

Few parameters you can change:

  • /pti/pti_configs/hyperparameters.py:
    • first_inv_type = 'w+' -> Use pretrained e4e encoder
    • first_inv_type = 'w' -> Use projection and optimization
  • /pti/pti_configs/paths_config.py:
    • input_data_path: path of real images
    • e4e: path of e4e_w+.pt
    • stylegan2_ada_shhq: pretrained stylegan2-ada model for SHHQ
python run_pti.py

Note: we used the test image under 'aligned_image/' (the output of alignment.py), the inverted latent code and fine-tuned generator will be saved in 'outputs/pti/'

Editing with InterfaceGAN, StyleSpace, and Sefa

python edit.py --network pretrained_models/stylegan_human_v2_1024.pkl --attr_name upper_length \\
    --seeds 61531,61570,61571,61610 --outdir outputs/edit_results

Editing using inverted latent code

python edit.py ---network outputs/pti/checkpoints/model_test.pkl --attr_name upper_length \\
    --outdir outputs/edit_results --real True --real_w_path outputs/pti/embeddings/test/PTI/test/0.pt --real_img_path aligned_image/test.png

Note:

  1. ''upper_length'' and ''bottom_length'' of ''attr_name'' are available for demo.
  2. Layers to control and editing strength are set in edit/edit_config.py.

Demo for InsetGAN

We implement a quick demo using the key idea from InsetGAN: combining the face generated by FFHQ with the human-body generated by our pretrained model, optimizing both face and body latent codes to get a coherent full-body image. Before running the script, you need to download the FFHQ face model, or you can use your own face model, as well as pretrained face landmark and pretrained CNN face detection model for dlib

python insetgan.py --body_network=pretrained_models/stylegan_human_v2_1024.pkl --face_network=pretrained_models/ffhq.pkl \\
    --body_seed=82 --face_seed=43  --trunc=0.6 --outdir=outputs/insetgan/ --video 1 

Results

Editing

InsetGAN re-implementation

Editing with inverted real image

(from left to right: real image | inverted image | InterFaceGAN result | StyleSpace result | SeFa result)

upper_length_test.mp4

For more demo, please visit our web page .

TODO List

  • Release 1024x512 version of StyleGAN-Human based on StyleGAN3
  • Release 512x256 version of StyleGAN-Human based on StyleGAN1
  • Extension of downstream application (InsetGAN): Add face inversion interface to support fusing user face image and stylegen-human body image
  • Add Inversion Script into the provided editing pipeline
  • Release Dataset

Related Works

  • (ICCV 2023) UnitedHuman: Harnessing Multi-Source Data for High-Resolution Human Generation, Jianglin Fu et al. [Paper], [Code], [Project Page]
  • (SIGGRAPH 2022) Text2Human: Text-Driven Controllable Human Image Generation, Yuming Jiang et al. [Paper], [Code], [Project Page], [Dataset]
  • (ICCV 2021) Talk-to-Edit: Fine-Grained Facial Editing via Dialog, Yuming Jiang et al. [Paper], [Code], [Project Page], [Dataset]
  • (Technical Report 2022) Generalizable Neural Performer: Learning Robust Radiance Fields for Human Novel View Synthesis, Wei Cheng et al. [Paper], [Code], [Project Page], [Dataset]

Citation

If you find this work useful for your research, please consider citing our paper:

@article{fu2022styleganhuman,
      title={StyleGAN-Human: A Data-Centric Odyssey of Human Generation}, 
      author={Fu, Jianglin and Li, Shikai and Jiang, Yuming and Lin, Kwan-Yee and Qian, Chen and Loy, Chen-Change and Wu, Wayne and Liu, Ziwei},
      journal   = {arXiv preprint},
      volume    = {arXiv:2204.11823},
      year    = {2022}

Acknowlegement

Part of the code is borrowed from stylegan (tensorflow), stylegan2-ada (pytorch), stylegan3 (pytorch).

Popular repositories Loading

  1. StyleGAN-Human StyleGAN-Human Public

    StyleGAN-Human: A Data-Centric Odyssey of Human Generation

    Python 1.2k 145

  2. StyleNeRF StyleNeRF Public

    Forked from facebookresearch/StyleNeRF

    This is the open source implementation of the ICLR2022 paper "StyleNeRF: A Style-based 3D-Aware Generator for High-resolution Image Synthesis"

    Python 2

  3. stylegan2-pytorch stylegan2-pytorch Public

    Forked from rosinality/stylegan2-pytorch

    Implementation of Analyzing and Improving the Image Quality of StyleGAN (StyleGAN 2) in PyTorch

    Python 1

  4. DualStyleGAN DualStyleGAN Public

    Forked from williamyang1991/DualStyleGAN

    [CVPR 2022] Pastiche Master: Exemplar-Based High-Resolution Portrait Style Transfer

    Jupyter Notebook 1

  5. gnr gnr Public

    Forked from generalizable-neural-performer/gnr

    Implementation of "Generalizable Neural Performer: Learning Robust Radiance Fields for Human Novel View Synthesis"

    Python 1

  6. stylegan-human.github.io stylegan-human.github.io Public

    JavaScript 2