Are you sure you want to delete this task? Once this task is deleted, it cannot be recovered.
You can not select more than 25 topics
Topics must start with a chinese character,a letter or number, can include dashes ('-') and can be up to 35 characters long.
steven413d
cb8f6e82d7
|
4 years ago | |
|---|---|---|
| .. | ||
| Data_preprocessing | 4 years ago | |
| .DS_Store | 4 years ago | |
| README.md | 4 years ago | |
| data_loader.py | 5 years ago | |
| main.py | 4 years ago | |
| model_utils.py | 4 years ago | |
| parameter.py | 4 years ago | |
| run.sh | 5 years ago | |
| run_test.sh | 5 years ago | |
| tester.py | 4 years ago | |
| trainer.py | 4 years ago | |
| unet.py | 4 years ago | |
| utils.py | 4 years ago | |
README.md
Face parsing
A Pytorch implementation face parsing model trained by CelebAMask-HQ
Dependencies
- Pytorch 0.4.1
- numpy
- Python3
- Pillow
- opencv-python
- tenseorboardX
Preprocessing
- Move the mask folder, the image folder, and
CelebA-HQ-to-CelebA-mapping.txt( remove 1st line in advance ) under./Data_preprocessing - Run
python g_mask.py - Run
python g_partition.pyto split train set and test set.
Training
- Run
bash run.sh #GPU_num
Well-trained model
- The model can be downloaded here.
- The model (
model.pth) should be put under./models/parsenet - Mask labels are defined as following:
| Label list | ||
|---|---|---|
| 0: 'background' | 1: 'skin' | 2: 'nose' |
| 3: 'eye_g' | 4: 'l_eye' | 5: 'r_eye' |
| 6: 'l_brow' | 7: 'r_brow' | 8: 'l_ear' |
| 9: 'r_ear' | 10: 'mouth' | 11: 'u_lip' |
| 12: 'l_lip' | 13: 'hair' | 14: 'hat' |
| 15: 'ear_r' | 16: 'neck_l' | 17: 'neck' |
| 18: 'cloth' |
- Overall Per-pixel Acc: 93.42 ( train and evaluate according to CelebA train/test split )
Testing & Color visualization
- Run
bash run_test.sh #GPU_num - Results will be saved in
./test_results - Color visualized results will be saved in
./test_color_visualize - Another way for color visualization without using GPU: Run
python ./Data_preprocessing/g_color.py
No Description
Python other