Based on the PyTorch implementation.
The model should be trainable, but so far it's only tested for inference with pre-trained weights (included in Demo/weights
).
Run demo application to apply styles to jpeg images:
swift run FastStyleTransferDemo --style=candy --image=FastStyleTransfer/Demo/examples/cat.jpg --output=candy_cat.jpg
swift run FastStyleTransferDemo --style=mosaic --image=FastStyleTransfer/Demo/examples/cat.jpg --output=mosaic_cat.jpg
Run demo notebook in Colab!