AIParsing

This respository includes a PyTorch implementation of the TIP2022 paper AIParsing:Anchor-Free Instance-Level Human Parsing.

Requirements:

python 3.7
PyTorch 1.7.1
cuda 10.1

The detail environment can be find in AIParsing_env.yaml.

Compiling

Apex install:

git clone https://github.com/NVIDIA/apex
cd apex
pip install -v --disable-pip-version-check --no-cache-dir --global-option="--cpp_ext" --global-option="--cuda_ext" ./
or python setup.py build develop

cd models
sh make.sh

cd ..
python setup.py build develop

Dataset and pretrained model

Plesae download CIHP dataset

Well-trained models on the CIHP and LV-MHP datasets (MM:aiparsing)

Evaluation

bash test_CIHP_R50_75epoch.sh

Training

bash train_CIHP_R50_75epoch.sh

Acknowledgment

This project is created based on the Parsing R-CNN, CenterMask

If this code is helpful for your research, please cite the following paper:

@article{AIParsing2022,
  title={AIParsing: Anchor-Free Instance-Level Human Parsing},
  author={Sanyi Zhang, Xiaochun Cao, Guo-jun Qi, Zhanjie Song, Jie Zhou},
  journal={IEEE Transactions on Image Processing (TIP)},
  year={2022},
  volume={31},
  pages={5599-5612}
}