Implementation and enhancement of the model "Learning Graph Neural Networks for Image Style Transfer"
This project is an simple Pytorch implementation of the paper "Learning Graph Neural Networks for Image Style Transfer"
Blocks of code that we have implemented :
- Encoder : VGG19
- Image2patch
- Knn graph constructor
- GATv2
- Patch2feat
- AdaIN
- Pyramid feature
- Using GATv2 instead of GAT since the graph structure is similar to graph structure of “SYNTHETIC BENCHMARK: DICTIONARY LOOKUP” in GATv2.
- We define a threshold to decide to construct an edge or not instead of KNN.
- Improving Patch2Feat with Feature Pyramid network.
- The operation proposed required a tremendous amount of GPU mem
- Even though our implementation didn’t get the desired results, We think that if we have enough GPU for smaller patch stride the output would be better.
- Invest on GPU power
- Finding improvement in order to reduce the memory requirements (different encoder).
- Implement deformable graph convolutional operation