Plain folding

Dependencies

General

• Python 3
• PIL (pillow)
• NumPy (numpy)

Dataset visualization

• Matplotlib (matplotlib)
• CV2 (opencv-python)

Pix2Pix

• torch
• CV2 (opencv-python)

Data examples*

*generated images with inverted colors and black contour

Train Pix2Pix model

generate folding dataset:

python gen_pix2pix_dataset.py

move unfolding dataset into pix2pix folder

mv unfolding_1000_fc_3_af_0p2 pytorch-CycleGAN-and-pix2pix/datasets/unfolding_1000_fc_3_af_0p2/

start model training:

cd  pytorch-CycleGAN-and-pix2pix
python train.py --dataroot ./datasets/unfolding_1000_fc_3_af_0p2 --name pix2pix_unfolding_1000_fc_3_af_0p2 --model pix2pix --direction BtoA
python train.py --dataroot ./datasets/unfolding_1000_fc_3_af_0p2 --name cycle_gan_unfolding_1000_fc_3_af_0p2 --model cycle_gan

eval on real img data:

python test_on_real_imgs.py --dataroot ./datasets/unfolding_1000_fc_3_af_0p2 --name pix2pix_unfolding_1000_fc_3_af_0p2 --model pix2pix 
python test_on_real_imgs.py --dataroot ./datasets/unfolding_1000_fc_3_af_0p2 --name cycle_gan_unfolding_1000_fc_3_af_0p2 --model cycle_gan 

Train U-Net model

U-Net training