The project intends to implement the original paper on Deep Photo Style Transfer by Luan. et. al. We have built a pytorch implementation of the original model built by the authors.
- We have modified and built upon the publicly available pytorch implementation of Neural Style Transfer by Gatys. Et. al
- We have used the torch implementation of Matting Laplacian available on github.
Our model makes the following assumptions:
- Pretrained VGG19 as feature extractor for the images
- Conv4_2 used for content loss and Conv1_1, Conv2_1, Conv3_1, Conv4_1, Conv5_1 used for style losses.
- The input image is already photorealistic.
- Weights as mentioned in the paper.
Luan, Fujun, et al. "Deep photo style transfer." Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. 2017.
https://pytorch.org/tutorials/advanced/neural_style_tutorial.html
https://github.com/MarcoForte/closed-form-matting
https://github.com/LouieYang/deep-photo-styletransfer-tf
Vaishnavi Kotturu
Khushi Gupta