An out-of-box human parsing representation extractor.
Our solution ranks 1st for all human parsing tracks (including single, multiple and video) in the third LIP challenge!
Features:
- Out-of-box human parsing extractor for other downstream applications.
- Pretrained model on three popular single person human parsing datasets.
- Training and inferecne code.
- Simple yet effective extension on multi-person and video human parsing tasks.
conda env create -f environment.yaml
conda activate schp
pip install -r requirements.txt
IntegratedPIFu uses this extractor to get the pseudo groundtruth human parsing maps.
Pascal-Person-Part (exp-schp-201908270938-pascal-person-part.pth)
-
mIoU on Pascal-Person-Part validation: 71.46 %.
-
Pascal Person Part is a tiny single person human parsing dataset with 3000+ images. This dataset focus more on body parts segmentation. Pascal Person Part has 7 labels, including 'Background', 'Head', 'Torso', 'Upper Arms', 'Lower Arms', 'Upper Legs', 'Lower Legs'.
Make a checkpoints directory: mkdir checkpoints
Then, download the trained model (trained on the pascal dataset) from https://drive.google.com/uc?id=1E5YwNKW2VOEayK9mWCS3Kpsxf-3z04ZE
Put the downloaded file (exp-schp-201908270938-pascal-person-part.pth) into the checkpoints folder.
python simple_extractor.py --integratedpifu-dir [PATH_TO_IntegratedPIFu Folder]
Please cite our work if you find this repo useful in your research.
@article{li2020self,
title={Self-Correction for Human Parsing},
author={Li, Peike and Xu, Yunqiu and Wei, Yunchao and Yang, Yi},
journal={IEEE Transactions on Pattern Analysis and Machine Intelligence},
year={2020},
doi={10.1109/TPAMI.2020.3048039}}- Source Image.
- LIP Parsing Result.
- ATR Parsing Result.
- Pascal-Person-Part Parsing Result.
- Source Image.
- Instance Human Mask.
- Global Human Parsing Result.
- Multiple Human Parsing Result.
Our code adopts the InplaceSyncBN to save gpu memory cost.
There is also a PaddlePaddle Implementation of this project.