1. facades
2. edges2shoes
3. edges2handbags
4. apple2orange
5. zebra2horse

• dataset link: https://people.eecs.berkeley.edu/~tinghuiz/projects/pix2pix/datasets/
• dataset link: https://people.eecs.berkeley.edu/~taesung_park/CycleGAN/datasets/

pytorch implementaion(iteration = 8950,batch_size 200)
A	AB	ABA
B	BA	BAB

pytorch implementaion(iteration = 173850,batch_size= 200)
A	AB	ABA
B	BA	BAB

pytorch implementaion(iteration = 15450,batch_size= 200)
A	AB	ABA
B	BA	BAB

Using Pix2pix's unet and combination of LSGAN and L1 For Lconst loss

Can train some dataset for 256 x 256 image size(discogan_pytorch_gpu_torloader_deep.py)

pytorch implementaion(iteration = 23916,batch_size= 25)
A	AB	ABA
B	BA	BAB

pytorch implementaion(epoch = 106,batch_size= 25)
A	AB	ABA
B	BA	BAB

pytorch implementaion(epoch = 200,batch_size= 25)
A	AB	ABA
B	BA	BAB

pytorch implementaion(epoch = 299,batch_size= 25)
A	AB	ABA
B	BA	BAB

My codes are heavily rely on https://github.com/carpedm20/DiscoGAN-pytorch Special thanks to carpedm20

1. Isola, Phillip, et al. "Image-to-image translation with conditional adversarial networks." (Full paper: https://arxiv.org/pdf/1611.07004.pdf) Unpaired Image-to-Image Translation using Cycle-Consistent Adversarial Networks (arXiv:1703.10593 [cs.CV])
2. Learning to Discover Cross-Domain Relations with Generative Adversarial Networks (arXiv:1703.05192 [cs.CV])
3. DualGAN: Unsupervised Dual Learning for Image-to-Image Translation (arXiv:1704.02510 [cs.CV])
4. https://github.com/SKTBrain/DiscoGAN
5. https://github.com/carpedm20/DiscoGAN-pytorch
6. https://github.com/duxingren14/DualGAN
7. https://github.com/junyanz/CycleGAN
8. https://github.com/znxlwm/pytorch-CycleGAN
9. https://github.com/togheppi/DualGAN