Skip to content
A configurable tf.keras based neural network image styler
Python
Branch: master
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Permalink
Type Name Latest commit message Commit time
Failed to load latest commit information.
assets
neural_image_styler
.gitignore
LICENSE
README.md
requirements.txt
setup.py

README.md

Neural Image Styler

A configurable tf.keras based neural network image styler. Based on this tensorflow tutorial

Command Line

usage: neural_image_styler [-h] --input INPUT --style_reference
                           STYLE_REFERENCE --output OUTPUT
                           [--iterations ITERATIONS]
                           [--content_weight CONTENT_WEIGHT]
                           [--style_weight STYLE_WEIGHT]
                           [--show_plots SHOW_PLOTS]

optional arguments:
  -h, --help            show this help message and exit
  --input INPUT         image file to be processed
  --style_reference STYLE_REFERENCE
                        image file to extract style from
  --output OUTPUT       output image file save location
  --iterations ITERATIONS
                        number of fit iterations
  --content_weight CONTENT_WEIGHT
  --style_weight STYLE_WEIGHT
  --show_plots SHOW_PLOTS
                        show debug plots

Colab

This package can be experimented with in a colab environment

Cloud Colab

See this Colab for a runnable colab, be sure to use a GPU runtime

Local Colab

See this guide for running a local Colab runtime to utilize local gpu resources.

Example

The output image is the content image stylized by the style image. result

As more iterations are completed, the output begins to take shape. iterations

The final result has the lowest loss. final

You can’t perform that action at this time.