This the official code for D2FP: Learning Implicit Prior for Human Parsing.

• Paper

This codebase is implemented using Detectron2.

• Linux or macOS with Python ≥ 3.6
• PyTorch ≥ 1.9 and torchvision that matches the PyTorch installation. Install them together at pytorch.org to make sure of this. Note, please check PyTorch version matches that is required by Detectron2.
• Detectron2: follow Detectron2 installation instructions.
• OpenCV is optional but needed by demo and visualization

After preparing the required environment, run the following command to compile CUDA kernel for MSDeformAttn:

CUDA_HOME must be defined and points to the directory of the installed CUDA toolkit.

cd m2fp/modeling/pixel_decoder/ops
sh make.sh

To build on a system that does not have a GPU device but provide the drivers:

TORCH_CUDA_ARCH_LIST='8.0' FORCE_CUDA=1 python setup.py build install

Download the R-103.pkl weights from the official website https://github.com/facebookresearch/MaskFormer/blob/main/MODEL_ZOO.md and place it in D2FP/weights/R-103.pkl

Set the DETECTRON2_DATASETS environment variable and organize the downloaded datasets in the following directory structure within the specified path. You can download the dataset from the official website: https://sysu-hcp.net/lip/overview.php.

lip/
  Training/
    Images/             # all training images
    Category_ids/       # all training category labels
  Validation/
    Images/
    Category_ids/

cihp/
  Training/
    Images/             # all training images
    Category_ids/       # all training category labels
    Instance_ids/       # all training part instance labels
    Human_ids/          # all training human instance labels
  Validation/
    Images/
    Category_ids/
    Instance_ids/
    Human_ids/

Set the number of GPUs and customize the configuration as needed.

./train.sh

Set the number of GPUs and customize the configuration as needed.

./test.sh

Code is largely based on M2FP (https://github.com/soeaver/M2FP).