Group 2 - Ethnicity Cycle GAN

Used articles and papers

1. Unpaired Image-to-Image Translation using Cycle-Consistent Adversarial Networks
   Implementation: https://github.com/junyanz/pytorch-CycleGAN-and-pix2pix
   Paper: https://arxiv.org/pdf/1703.10593.pdf

2. Article having (1.) as basis

https://hackernoon.com/gender-and-race-change-on-your-selfie-with-neural-nets-9a9a1c9c5c16

Goal
Having Cycle GANs for ethnicity transformation, e.g.

1. Black and White

1. White and Asian

Used Datasets

1. CelebA ~ 200.000 images
   http://mmlab.ie.cuhk.edu.hk/projects/CelebA.html

2. UTKFace ~ 20.000 images
   http://aicip.eecs.utk.edu/wiki/UTKFace

3. LFW (Labeled Faces in the Wild) Database ~ 13.000 images http://vis-www.cs.umass.edu/lfw/

Faced Problems

1. CelebA is a huge dataset but does not have ethnicity labels unfortunately.
   Approach: train Classifier on UTKFace and LFW to label CelebA.

Improvements 2. Replace L1 pixel to pixel identity loss with p2 Norm of feature map distance. see https://cs.stanford.edu/people/jcjohns/papers/eccv16/JohnsonECCV16.pdf