| Idx | Tasks | Done |
|---|---|---|
| 1 | CUB 11 Dataset setting | July 29.2019 |
| 2 | Text Enbedding using Google Bert | July 29.2019 |
| 3 | Dataset.py Transforms.py combine | July 30.2019 |
| 4 | Train.py model.py C&debug | July 31.2019 |
| 5 | Run the test script (only the vanilla GAN) successfully | Aug 02.2019 |
| 6 | Run the GAWWN script successfully |
- Get dataset here : (CUB) http://www.vision.caltech.edu/visipedia/CUB-200-2011.html
- Bert_preprocessing.py to create the Bert text encoding
- modify the config.cfg
- python3 GAWWN_train_img_only.py for vanilla GAN training
- python3 GAWWN_train.py for GAWWN training (Not yet completed)
| Types | images |
|---|---|
| vanilla GAN v1 basic structure from mnist test (epoch 100) |  |
| vanilla GAN v1 basic structure from mnist test (epoch 200) |  |
| vanilla GAN v2 add LeakyReLU, BN, deep structure (epoch 100, 64 * 64) |  |
| vanilla GAN v2 add LeakyReLU, BN, deep structure (epoch 200, 64 * 64) |  |
| vanilla GAN v2 add LeakyReLU, BN, deep structure (epoch 300, 64 * 64) |  |
| vanilla GAN v2 add LeakyReLU, BN, deep structure (epoch 300, 128 * 128) |  |
| vanilla GAN v2 add LeakyReLU, BN, deep structure (epoch 300, 128 * 128) |  |
| vanilla GAN v2 add LeakyReLU, BN, deep structure (epoch 300, 128 * 128) |  |
| vanilla GAN v3 add balanced G&D structure (epoch 300, 128 * 128) |  |
| vanilla GAN v3 add balanced G&D structure and soft BCELoss (epoch 300, 128 * 128) |  |