Implementation of Neural Style Transfer on Video.
Contributers : Fenil Doshi, Rohan Pooniwala, Rohit Saha, Raghav Gupta
We mainly used the following research papers to implement Neural Artistic Style
- Implementation of Neural Style Transfer from the paper A Neural Algorithm of Artistic Style in Keras 1.2.0
- Color Preservation is based on the paper [Preserving Color in Neural Artistic Style Transfer](https://arxiv.org/abs/1606.05897).
- Masked Style Transfer is based on the paper [Show, Divide and Neural: Weighted Style Transfer](http://cs231n.stanford.edu/reports2016/208_Report.pdf)
And also took help from the repository Neural-Style-Transfer
We are using Keras 1.2.1 with Tensorflow backend Using CUDA on Windows 10. But it should work in Linux with/without CUDA using same keras version.
On a 1070 Laptop GPU, the time required for each epoch depends on mainly image size (gram matrix size) :
- For a 400x400 gram matrix, each epoch takes approximately 7-9 seconds.
- For a 512x512 gram matrix, each epoch takes approximately 15-16 seconds.
- For a 600x600 gram matrix, each epoch takes approximately 23-25 seconds.