Awesome

Conditional GAN

Conditional Generative Adversarial Networks for anime generation (AnimeGAN).

<br /> Training results dump every 500 min-batch in 25 epoch(26000th min-batch) for the following tags

• blue hair blue eyes <br />
• gray hair green eyes <br />
• green hair red eyes <br />
• orange hair brown eyes <br />
• blonde hair gray eyes <br />
• pink hair aqua eyes <br />

Sample training data

<br />

Environment

python3 <br /> tensorflow 1.0 <br /> scipy <br />

Model structure

Data

source link <br /> google drive link

Usage

1. Download hw3 data from data link, place the MLDS_HW3_dataset/ in the same directory and unzip the face.zip in MLDS_HW3_dataset/
2. Replace the tags in MLDS_HW3_dataset/sample_testing_text.txt to the right format.
3. Start training !

Train

First time use, you need to do the preprocessing

$ python3 main.py --prepro 1

If you already have done the preprocessing

$ python3 main.py --prepro 0

Model

• dcgan structure
• use one hot encoding for condition tags

Test

This code will automatically dump the results for the tags specified in MLDS_HW3_dataset/sample_testing_text.txt every <em>dump_every</em> batches to the test_img/ folder. <br />

Testing tags format

1,<Color> hair <Color> eyes 
2,<Color> hair <Color> eyes
3,<Color> hair <Color> eyes
4,<Color> hair <Color> eyes
.
.
.

• Possible colors for eyes

['<UNK>', 'yellow', 'gray', 'blue', 'brown', 'red', 'green', 'purple', 'orange',
 'black', 'aqua', 'pink', 'bicolored']

• Possible colors for hair

['<UNK>', 'gray', 'blue', 'brown', 'red', 'blonde', 'green', 'purple', 'orange',
 'black', 'aqua', 'pink', 'white']