https://github.com/sqmah/art-style-transfer-video

Art style transfer for both images and video written in Tensorflow 2. Preview and save your outputs.
https://github.com/sqmah/art-style-transfer-video

gan style-transfer tensorflow tensorflow-examples tensorflow2

Last synced: 4 months ago
JSON representation

Art style transfer for both images and video written in Tensorflow 2. Preview and save your outputs.

• Host: GitHub
• URL: https://github.com/sqmah/art-style-transfer-video
• Owner: SQMah
• Created: 2021-04-25T09:09:53.000Z (almost 4 years ago)
• Default Branch: main
• Last Pushed: 2021-10-08T20:10:20.000Z (over 3 years ago)
• Last Synced: 2024-10-10T09:01:03.291Z (4 months ago)
• Topics: gan, style-transfer, tensorflow, tensorflow-examples, tensorflow2
• Language: Python
• Homepage:
• Size: 70.8 MB
• Stars: 4
• Watchers: 1
• Forks: 2
• Open Issues: 0
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

README

        
# Art-Style-Transfer-Video

Art style transfer for images and *video*. Save and resize your outputs. Example artworks included.



## Setup

Use your choice of Python 3.7+ virtual environment or conda environment.

Then run:

``pip install -r requirements.txt``

## Examples

Find example images to transfer art styles from in ``art``.

## Video

``python video.py``

Note the following arguments, which are all optional:

```

usage: video.py [-h] [--i I] [--s S] [--u U] [--r R] [--w W] [--f F]

optional arguments:

  -h, --help            show this help message and exit

  --i I, ---img_name I  Image name of the artwork to transfer the style from

                        in the art folder.

  --s S, --src S        Input source. Defaults to 0 for web cam.

  --u U, --url U        URL to the tf hub model.

  --r R, --resolution R

                        Resolution of the smallest dimension of the input.

  --w W, --write W      Whether or not to write the output.

  --f F, --frame_skip F

                        Number of frames to skip between processing.

                        

```

## Image

``python img.py [path to input image]``

Pass any image preprocessing logic (i.e. blurring) into the ``preprocess`` argument as a function. Note that the preprocessing function should return a ``numpy.ndarray``.

Note the following arguments, which are all optional except for src:

```

usage: img.py [-h] [--i I] [--u U] [--r R] [--w W] src

positional arguments:

  src                   Input image

optional arguments:

  -h, --help            show this help message and exit

  --i I, ---img_name I  Image name of the artwork to transfer the style from

                        in the art folder.

  --u U, --url U        URL to the tf hub model.

  --r R, --resolution R

                        Resolution of the smallest dimension of the input.

  --w W, --write W      Whether or not to write the output.

```