InceptionScore

Inheritance Diagram

Inheritance diagram of ashpy.metrics.gan.InceptionScore

class ashpy.metrics.gan.InceptionScore(inception, model_selection_operator=<built-in function gt>, logdir='/home/docs/checkouts/readthedocs.org/user_builds/ashpy/checkouts/v0.2.0/docs/source/log')[source]

Bases: ashpy.metrics.metric.Metric

Inception Score Metric.

This class is an implementation of the Inception Score technique for evaluating a GAN.

See Improved Techniques for Training GANs 1.

1

Improved Techniques for Training GANs https://arxiv.org/abs/1606.03498

Methods

__init__(inception[, …])

Initialize the Metric.

get_or_train_inception(dataset, name, …[, …])

Restore or train (and save) the Inception model.

inception_score(images)

Compute the Inception Score.

update_state(context)

Update the internal state of the metric, using the information from the context object.

Attributes

best_folder

Retrieve the folder used to save the best model when doing model selection.

best_model_sel_file

Retrieve the path to JSON file containing the measured performance of the best model.

logdir

Retrieve the log directory.

metric

Retrieve the tf.keras.metrics.Metric object.

model_selection_operator

Retrieve the operator used for model selection.

name

Retrieve the metric name.

__init__(inception, model_selection_operator=<built-in function gt>, logdir='/home/docs/checkouts/readthedocs.org/user_builds/ashpy/checkouts/v0.2.0/docs/source/log')[source]

Initialize the Metric.

Parameters
  • inception (tf.keras.Model) – Keras Inception model.

  • model_selection_operator (typing.Callable) –

    The operation that will be used when model_selection is triggered to compare the metrics, used by the update_state. Any typing.Callable behaving like an operator is accepted.

    Note

    Model selection is done ONLY if an operator is specified here.

  • logdir (str) – Path to the log dir, defaults to a log folder in the current directory.

static get_or_train_inception(dataset, name, num_classes, epochs, fine_tuning=False, loss_fn=<tensorflow.python.keras.losses.SparseCategoricalCrossentropy object>, optimizer=<tensorflow.python.keras.optimizer_v2.adam.Adam object>, logdir='/home/docs/checkouts/readthedocs.org/user_builds/ashpy/checkouts/v0.2.0/docs/source/log')[source]

Restore or train (and save) the Inception model.

Parameters
  • dataset (tf.data.Dataset) – Dataset to re-train Inception Model on.

  • name (str) – Name of this new Inception Model, used for saving it.

  • num_classes (int) – Number of classes to use for classification.

  • epochs (int) – Epochs to train the Inception model for.

  • fine_tuning (bool) – Controls wether the model will be fine-tuned or used as is.

  • loss_fn (tf.keras.losses.Loss) – Keras Loss for the model.

  • optimizer (tf.keras.optimizers.Optimizer) – Keras optimizer for the model.

  • logdir (str) – Path to the log dir, defaults to a log folder in the current directory.

Return type

Model

Returns

tf.keras.Model – The Inception Model.

inception_score(images)[source]

Compute the Inception Score.

Parameters

images (list of [numpy.ndarray]) – A list of ndarray of generated images of 299x299 of size.

Return type

Tensor

Returns

tuple of (numpy.ndarray, numpy.ndarray) – Mean and STD.

update_state(context)[source]

Update the internal state of the metric, using the information from the context object.

Parameters

context (ashpy.contexts.ClassifierContext) – An AshPy Context holding all the information the Metric needs.

Return type

None