The algorithm is described in the ICCV 2019 paper: Learning Compositional Neural Infomation Fusion for Human Parsing. In this work, we propose to combine neural networks with the compositional hierarchy of human bodies for efficient and complete human parsing.

This repository is developed under CUDA9.0 and pytorch-0.4.1 in python3.6. Early versions of pytorch can be found here. The required packages can be installed by:

pip install -r requirements.txt

$CompositionalHumanParsing
├── dataset
│   ├── list
├── doc
├── inplace_abn
│   ├── src
├── module
├── network
├── progress

1. Download the pre-trained model.

python evaluate.py --root <path to the dataset> --restore_from <path to the pre-traiend model>

If you find this code useful, please cite our work with the following bibtex:

@inproceedings{wang2019learning,
  title={Learning compositional neural information fusion for human parsing},
  author={Wang, Wenguan and Zhang, Zhijie and Qi, Siyuan and Shen, Jianbing and Pang, Yanwei and Shao, Ling},
  booktitle={Proceedings of the IEEE International Conference on Computer Vision},
  pages={5703--5713},
  year={2019}
}