This project was an assignment for the course "Advance Deep Learning for Computer Graphics" at University of Bonn, Summer Semester 2020
- PyTorch
- TorchVision
- Numpy
- Scipy
- Training:
python neural_style/neural_style.py train --dataset </path/to/train-dataset> --style-image </path/to/style/image> --save-model-dir </path/to/save-model/folder> --epochs 2 --cuda 1
- Evaluation on an Image
python neural_style/neural_style.py eval --content-image </path/to/content/image> --model </path/to/saved/model> --output-image </path/to/output/image> --cuda 0
- epochs - # of epochs to train (2 is more than enough)
- cuda - 0 for cpu, 1 for gpu usage
Path to download data : http://mscoco.org/dataset/#download. (13 Gb)