paper

• Data: WikiArt
  now dataset

2017-10-06

https://drive.google.com/file/d/0B7YClE2-vjgPbGE3LUd5SUQ3Ukk/view?usp=sharing

• Gram Martix
• Self-Organizing maps(SOM)

Reference:

1. style classification and visualization of art painting's genre using self-organizing maps
2. image style transfer using convolutional neural networks
3. SOM introduce
4. SOM code

2017-10-13

• Visualizing CNN
• How to combine Gram and SOM ? use 'PCA' to 2048 or 4096

Reference:

1. style transfer with Keras
2. conv filter visualization with Keras
3. conv class activation maps with Keras

2017-10-20

https://drive.google.com/file/d/0B7YClE2-vjgPUEd1VmF0Wk1oSm8/view?usp=sharing

• Improve 'feature(Gram)'
• Implementation

Reference:

1. Deep Correlation Features for Image Style Classification

2017-10-28

https://drive.google.com/file/d/0B7YClE2-vjgPU0ctbXFNc1N2Z0k/view?usp=sharing

• Implementation

2017-11-03

paper: https://drive.google.com/file/d/0B7YClE2-vjgPR3VYOUFmRnRkOWM/view?usp=sharing https://drive.google.com/file/d/0B7YClE2-vjgPOTlJYW1QRHZpeTA/view?usp=sharing

• (keras) images->VGG16->Gram->PCA->SOM

2017-11-10

• Goal
• VGG + PCA/Autoencoder
• VGG + PCA/Autoencoder + SOM
• VGG->Gram + PCA/Autoencoder

2017-11-17

https://drive.google.com/file/d/1nrX-ZC4bqiTYVhQ07Ro1uYqnI93VKCJF/view?usp=sharing