Skip to content

Releases: Xzzit/fast-style-transfer

v 1.0.1

13 Jun 01:05
Compare
Choose a tag to compare

Fix a bug for cpu and mac m1 user.
Update readme

v 1.0.0 fast style transfer

08 Jun 10:18
Compare
Choose a tag to compare

It is the stable version of fast style transfer!

What is different between this one and other fast style transfer repo:

  1. It implements state-of-the-art CNN architecture for stylizing given content images.
  2. It adds two additional loss items for more stable output: consistency loss and tv loss.
  3. The size of the style transfer neural network is incredibly small, with only 7 MB.
  4. The code supports the latest PyTorch version (aka. PyTorch 2.0).
  5. This code uses a command line argument as input, which is easier and more friendly for those who are new to Python.