• head swap
• change cloth
• generate neck & arms considering skin color.

(If you want to show more detail, go to etc folder.)

We used a MVC dataset. Please download this first.

• Resize

Image's original size is (1920, 2240), but this is too big to learn for computer. So, we need to resize this.

Before run this code, you must be set TARGET_SIZE and base_dir.

python resize.py

• Segmentation

Use opensource for segmentation. You must save images in 'p','1','2',3',... in image folder.

We use CIHP_PGN.

(128,0,0),(255,0,0),(0,85,0),(170,0,51),(255,85,0)
1=Hat, 2=Hair, 3=Glove, 4=Sunglasses, 5=UpperClothes
(0,0,85),(0,119,221),(85,85,0),(0,85,85),(85,51,0)
6=Dress, 7=Coat, 8=Socks, 9=Pants, 10=Neck
(52,86,128),(0,128,0),(0,0,255),(51,170,221),(0,255,255)
11=Scarf, 12=Skirt, 13=Face, 14=LeftArm, 15=RightArm
(85,255,170),(170,255,85),(255,255,0),(255,170,0)]
16=LeftLeg, 17=RightLeg, 18=LeftShoe, 19=RightShoe

• Pose

Use opensource for pose. You must save images in 'p','1','2',3',... in image folder.

We use pytorch_Realtime_Multi-Person_Pose_Estimation.

• Crop

Crop clothes from person's image.

Before run this code, you must be set TARGET_SIZE, IS_TOPS, base_dir. The meaning of IS_TOPS is whether you are learning tops or bottoms.

python crop.py

• Split

Split train data and test data.

Before run this code, you must be set base_dir.

python stage0/split.py

• Make pair txt

Make pair in same clothes image.

Before run this code, you must be set base_dir.

python mkpairtxt.py

It is in utils.py.

• PWD

  • your repository's full path
  • str(ex : '/home/fashionteam/ClothFlow/')

• IS_TOPS

  • whether to wear tops or bottoms
  • bool(ex : True)

• REAL_TEST

  • test for training or not
  • bool(ex : True)

• TENSORBOARD

  • how to display for you(tensorboard or console log)
  • bool(ex : True)

Change condition cloth mask to target cloth mask

• Train

python stage1/train.py

• Warped_Mask

python stage1/test.py

-> result/warped_mask & test/warped_mask

Result

Warp

• Train

We need a warped_mask folder.

python stage2/train.py

• Warped_Cloth

python stage2/test.py

-> result/warped_cloth & test/warped_cloth Result

wear cloth

• Train

We need warped_mask, warped_cloth folders.

python stage3/train.py

Result

• Final Result

python stage3/test.py

-> result/final & test/final

python inference.py