Class ModelLoader

Barracuda Model loader

Inheritance

Object

ModelLoader

Inherited Members

Object.ToString()

Object.Equals(Object)

Object.Equals(Object, Object)

Object.ReferenceEquals(Object, Object)

Object.GetHashCode()

Object.GetType()

Object.MemberwiseClone()

Namespace: Unity.Barracuda

Syntax

public static class ModelLoader

Methods

Load(Byte[], Boolean, Boolean)

Return an object oriented representation (aka: Model) of a neural network from a byte[] array. By default details are not logged to the console, set verbose to true to see loading details.

Declaration

public static Model Load(byte[] stream, bool verbose = false, bool skipWeights = false)

Parameters

Type	Name	Description
Byte[]	stream	binary representation of model as a byte array
Boolean	verbose	verbose
Boolean	skipWeights	skip loading weights (fast loading, metadata only)

Returns

Type	Description
Model	loaded Model

Load(String, Boolean, Boolean)

Return an object oriented representation (aka: Model) of a neural network from a .bc file. By default details are not logged to the console, set verbose to true to see loading details.

Declaration

public static Model Load(string filepath, bool verbose = false, bool skipWeights = false)

Parameters

Type	Name	Description
String	filepath	file name
Boolean	verbose	verbose
Boolean	skipWeights	skip loading weights (fast loading, metadata only)

Returns

Type	Description
Model	loaded Model

Load(NNModel, Boolean, Boolean)

Return an object oriented representation (aka: Model) of a neural network from a binary representation of type NNModel. By default details are not logged to the console, set verbose to true to see loading details.

Declaration

public static Model Load(NNModel model, bool verbose = false, bool skipWeights = false)

Parameters

Type	Name	Description
NNModel	model	model
Boolean	verbose	verbose
Boolean	skipWeights	skip loading weights (fast loading, metadata only)

Returns

Type	Description
Model	loaded Model

LoadAsync(Byte[], Model, Boolean, Boolean, Single)

Return an object oriented representation (aka: Model) of a neural network from a byte[] array. By default details are not logged to the console, set verbose to true to see loading details.

Declaration

public static IEnumerator LoadAsync(byte[] stream, Model model, bool verbose = false, bool skipWeights = false, float maxTimePerYield = 0.01F)

Parameters

Type	Name	Description
Byte[]	stream	binary representation of model as a byte array
Model	model	object-oriented representation of model (must initialize before calling method)
Boolean	verbose	verbose
Boolean	skipWeights	skip loading weights (fast loading, metadata only)
Single	maxTimePerYield	the maximum amount of time to spend between in computation before yielding

Returns

Type	Description
IEnumerator	IEnumerator (use with StartCoroutine)

LoadAsync(String, Model, Boolean, Boolean, Single)

Return an object oriented representation (aka: Model) of a neural network from a .bc file. By default details are not logged to the console, set verbose to true to see loading details.

Declaration

public static IEnumerator LoadAsync(string filepath, Model model, bool verbose = false, bool skipWeights = false, float maxTimePerYield = 0.01F)

Parameters

Type	Name	Description
String	filepath	file name
Model	model	object-oriented representation of model (must initialize before calling method)
Boolean	verbose	verbose
Boolean	skipWeights	skip loading weights (fast loading, metadata only)
Single	maxTimePerYield	the maximum amount of time to spend between in computation before yielding

Returns

Type	Description
IEnumerator	IEnumerator (use with StartCoroutine)

LoadAsync(NNModel, Model, Boolean, Boolean, Single)

Return an object oriented representation (aka: Model) of a neural network from a binary representation of type NNModel. By default details are not logged to the console, set verbose to true to see loading details.

Declaration

public static IEnumerator LoadAsync(NNModel nnModel, Model model, bool verbose = false, bool skipWeights = false, float maxTimePerYield = 0.01F)

Parameters

Type	Name	Description
NNModel	nnModel	binary representation of model
Model	model	object-oriented representation of model (must initialize before calling method)
Boolean	verbose	verbose
Boolean	skipWeights	skip loading weights (fast loading, metadata only)
Single	maxTimePerYield	the maximum amount of time to spend between in computation before yielding

Returns

Type	Description
IEnumerator	IEnumerator (use with StartCoroutine)

LoadAsyncFromStreamingAssets(String, Model, Boolean, Boolean, Single)

Return an object oriented representation (aka: Model) of a neural network from a .bc file from the the streaming asset folder. By default details are not logged to the console, set verbose to true to see loading details.

Declaration

public static IEnumerator LoadAsyncFromStreamingAssets(string filename, Model model, bool verbose = false, bool skipWeights = false, float maxTimePerYield = 0.01F)

Parameters

Type	Name	Description
String	filename	file name
Model	model	object-oriented representation of model (must initialize before calling method)
Boolean	verbose	verbose
Boolean	skipWeights	skip loading weights (fast loading, metadata only)
Single	maxTimePerYield	the maximum amount of time to spend between in computation before yielding

Returns

Type	Description
IEnumerator	IEnumerator (use with StartCoroutine)

LoadFromStreamingAssets(String, Boolean, Boolean)

Return an object oriented representation (aka: Model) of a neural network from a .bc file from the the streaming asset folder. By default details are not logged to the console, set verbose to true to see loading details.

Declaration

public static Model LoadFromStreamingAssets(string filename, bool verbose = false, bool skipWeights = false)

Parameters

Type	Name	Description
String	filename	file name
Boolean	verbose	verbose
Boolean	skipWeights	skip loading weights (fast loading, metadata only)

Returns

Type	Description
Model	loaded Model