API Reference

stylish

stylish.transform_image(path, style_path, output_path, vgg_path, iterations=None, learning_rate=None, content_weight=None, style_weight=None, tv_weight=None, content_layer=None, style_layers=None, log_path=None)[source]

Generate new image from path with style from another image.

Usage example:

>>> transform_image(
...    "/path/to/image.jpg",
...    "/path/to/style_image.jpg",
...    "/path/to/output_image/",
...    "/path/to/vgg_model.mat"
... )
Parameters:
  • path – path to the image to transform.
  • style_path – path to an image from which the style features will be extracted.
  • output_path – path where the transformed image will be generated.
  • vgg_path – path to the Vgg19 pre-trained model in the MatConvNet data format.
  • iterations – number of time that image should be trained against style_path. Default is stylish.core.ITERATIONS_NUMBER.
  • learning_rateLearning Rate value to train the model. Default is stylish.core.LEARNING_RATE.
  • content_weight – weight of the content feature cost. Default is stylish.core.CONTENT_WEIGHT.
  • style_weight – weight of the style feature cost. Default is stylish.core.STYLE_WEIGHT.
  • tv_weight – weight of the total variation cost. Default is stylish.core.TV_WEIGHT.
  • content_layer – Layer name from pre-trained Vgg19 model used to extract the content information. Default is stylish.vgg.CONTENT_LAYER.
  • style_layers – Layer names from pre-trained Vgg19 model used to extract the style information with corresponding weights. Default is stylish.vgg.STYLE_LAYERS.
  • log_path – path to extract the log information. Default is the same path as the output path.
Returns:

path to transformed image.

stylish.create_model(training_path, style_path, output_path, vgg_path, learning_rate=None, batch_size=None, batch_shape=None, epoch_number=None, content_weight=None, style_weight=None, tv_weight=None, content_layer=None, style_layers=None, limit_training=None, log_path=None)[source]

Train a style generator model based on an image and a dataset folder

Usage example:

>>> create_model(
...    "/path/to/training_data/",
...    "/path/to/style_image.jpg",
...    "/path/to/output_model/",
...    "/path/to/vgg_model.mat"
... )
Parameters:
  • training_path – training dataset folder.
  • style_path – path to an image from which the style features will be extracted.
  • output_path – path where the trained model and logs should be saved
  • vgg_path – path to the Vgg19 pre-trained model in the MatConvNet data format.
  • learning_rateLearning Rate value to train the model. Default is stylish.core.LEARNING_RATE.
  • batch_size – number of images to use in one training iteration. Default is stylish.core.BATCH_SIZE.
  • batch_shape – shape used for each images within training dataset. Default is stylish.core.BATCH_SHAPE.
  • epoch_number – number of time that model should be trained against training_images. Default is stylish.core.EPOCHS_NUMBER.
  • content_weight – weight of the content feature cost. Default is stylish.core.CONTENT_WEIGHT.
  • style_weight – weight of the style feature cost. Default is stylish.core.STYLE_WEIGHT.
  • tv_weight – weight of the total variation cost. Default is stylish.core.TV_WEIGHT.
  • content_layer – Layer name from pre-trained Vgg19 model used to extract the content information. Default is stylish.vgg.CONTENT_LAYER.
  • style_layers – Layer names from pre-trained Vgg19 model used to extract the style information with corresponding weights. Default is stylish.vgg.STYLE_LAYERS.
  • limit_training – maximum number of files to use from the training dataset folder. By default, all files from the training dataset folder are used.
  • log_path – path to extract the log information. Default is the same path as the output path.
Returns:

None

stylish.apply_model(model_path, input_path, output_path)[source]

Apply style generator model to a new image.

Usage example:

>>> apply_model(
...    "/path/to/saved_model/",
...    "/path/to/image.jpg",
...    "/path/to/output_image/"
... )

/path/to/output/image.jpg
Parameters:
  • model_path – path to trained model saved.
  • input_path – path to image to inferred model to.
  • output_path – path folder to save image output.
Returns:

path to transformed image.

stylish.extract_style_pattern(path, output_path, vgg_path, style_layers=None)[source]

Generate style pattern images from image path.

Usage example:

>>> apply_model(
...    "/path/to/style_image.jpg",
...    "/path/to/output/"
...    "/path/to/vgg_model.mat"
... )

/path/to/output/image.jpg
Parameters:
  • path – path to the image to extract style pattern images from.
  • output_path – path where the images will be generated.
  • vgg_path – path to the Vgg19 pre-trained model in the MatConvNet data format.
  • style_layers – Layer names from pre-trained Vgg19 model used to extract the style information with corresponding weights. Default is stylish.vgg.STYLE_LAYERS.
Returns:

list of image paths generated.