apple/ml-gmpi

Code for estimating pose for preprocessing (used in "Deep3DFaceRecon" directory)

Closed this issue · 1 comments

mjmjeong commented

In the TRAIN_EVAL.md, you used the following commands for estimating camera pose:

# For FFHQ
cd ${Deep3DFaceRecon_PATH}
python estimate_pose_ffhq.py --name=pretrained --epoch=20 --img_folder=${GMPI_ROOT}/runtime_dataset/dummy --gmpi_img_res ${RES} --gmpi_root ${GMPI_ROOT}

# For MetFace
cd ${Deep3DFaceRecon_PATH}
python estimate_pose_metfaces.py --name=pretrained --epoch=20 --img_folder=${GMPI_ROOT}/runtime_dataset/dummy --gmpi_root ${GMPI_ROOT}

However, I can not find these codes in the Deep3DFaceRecon_pytorch repository.
I guess you newly made these python files. Could you share these files?

Xiaoming-Zhao commented

Hi thanks for your interests. In the TRAIN_EVAL.md, I mentioned that using my modified version of the Deep3DFaceRecon_pytorch. I only modified some small things, e.g., adding the scripts you mentioned and adding several new command line arguments etc.

Maybe I should make it clearer.

  • 👍1