Skip to content

Latest commit

 

History

History
73 lines (71 loc) · 3.24 KB

tailornet_preparation.md

File metadata and controls

73 lines (71 loc) · 3.24 KB
Raw

TailorNet Data Preparation

  1. Download the SMPL models (Female, Male and Neutral) from this link (requires registration)
  2. Unzip the file and put the three *.pkl files from SMPL_python_v.1.1.0/smpl/models/ in a directory <tailornet_data_root>/smpl
    • <tailornet_data_root> can be choosen arbitrarily
  3. In garmentor/tailornet_dataset_for_garmentor/global_var.py , make sure that the filenames for SMPL_PATH_MALE and SMPL_PATH_FEMALE are the same as the filenames of the *.pkl files that you unzipped in step 2
    • The filenames are the last arguments in the os.path.join() calls
    • It's ok that no neutral model is listed here
  4. In garmentor/tailornet_for_garmentor/global_var.py , make sure that the filenames for SMPL_PATH_[FEMALE, MALE, NEUTRAL] correspond to the filenames of the *.pkl files that you unzipped in step 2
    • Only modify the filenames but keep the paths as they are
  5. Run 
    garmentor/setup_scripts/setup_tailornet.sh /absolute/path/to/garmentor/directory <absolute_tailornet_data_root_dir>
    • This converts the SMPL models for use in TailorNet
  6. Download TailorNet dataset from here
    • Download dataset_meta.zip (dataset meta data)
    • Download one or more sub-datasets (each represents a different garment type):
      • t-shirt_female(6.9 GB)
      • t-shirt_male(7.2 GB)
      • old-t-shirt_female(10 GB)
      • t-shirt_female_sample(19 MB)
      • shirt_female(12.7 GB)
      • shirt_male(13.5 GB)
      • pant_female(3.3 GB)
      • pant_male(3.4 GB)
      • short-pant_female(1.9 GB)
      • short-pant_male(2 GB)
      • skirt_female(5 GB)
  7. Unzip the downloaded files to <tailornet_data_root>
  8. For the garment types you downloaded, download the pre-trained weights from here
  9. Unzip the downloaded weights to <tailornet_data_root>/weights

Your <tailornet_data_root> should now look similar to this (depending on the different garment types that you downloaded):

<tailornet_data_root>/
├── apose.npy
├── apose.pkl
├── garment_class_info.pkl
├── garment_class_info_py2.pkl
├── pant_upper_boundary.npy
├── shirt_left_boundary.npy
├── shirt_right_boundary.npy
├── shirt_upper_boundary.npy
├── short-pant_upper_boundary.npy
├── skirt_upper_boundary.npy
├── skirt_weight.npz
├── smpl
│   ├── basicmodel_f_lbs_10_207_0_v1.1.0.pkl
│   ├── basicmodel_m_lbs_10_207_0_v1.1.0.pkl
│   ├── basicmodel_neutral_lbs_10_207_0_v1.1.0.pkl
│   ├── smpl_female.npz
│   ├── smpl_hres_female.npz
│   ├── smpl_hres_male.npz
│   ├── smpl_male.npz
├── split_static_pose_shape.npz
├── t-shirt_male
│   ├── avail.txt
│   ├── pivots.txt
│   ├── pose
│   ├── shape
│   ├── style
│   ├── style_model.npz
│   ├── style_shape
│   └── test.txt
├── t-shirt_upper_boundary.npy
└── weights
    └── t-shirt_male_weights/
           ...
    └── pant_male_weights/
           ...
    └── ...