AiliaFeatureExtractorModel Class Reference

Inheritance diagram for AiliaFeatureExtractorModel:

Collaboration diagram for AiliaFeatureExtractorModel:

Public Member Functions
bool	Settings (uint set_format, uint set_channel, uint set_range, uint set_distance_type, string set_layer_name)
	Model setting. More...
override bool	OpenFile (string prototxt, string model_path)
	Create a network object from a model file. More...
override bool	OpenEx (Ailia.ailiaFileCallback callback, IntPtr arg1, IntPtr arg2)
	Creates a network object from a file callback. More...
override bool	OpenMem (byte[] prototxt_buf, byte[] model_buf)
	Creates network objects from memory. More...
float []	ComputeFromImage (Color32 [] image, int image_width, int image_height)
	Acquire features from images. More...
float []	ComputeFromImageB2T (Color32 [] image, int image_width, int image_height)
	Obtain features from the upside-down image. More...
float	Match (float [] feature1, float [] feature2)
	Calculates the distance between features. More...
override void	Close ()
	Destroy feature extraction quobjects. More...
override void	Dispose ()
	Release resources. More...
Public Member Functions inherited from AiliaModel
bool	Environment (int type)
	Selects the specified type of calculation environment. More...
string	EnvironmentName ()
	Displays the name of the selected environment. More...
int	GetEnvironmentCount ()
	Get the number of available computing environments (CPU, GPU). More...
Ailia.AILIAEnvironment	GetEnvironment (int idx)
	Obtains the computing environment for the specified index. More...
bool	SelectEnvironment (int idx)
	Select the calculation environment. More...
Ailia.AILIAEnvironment	GetSelectedEnvironment ()
	Retrieves the selected computing environment. More...
void	SetMemoryMode (uint set_memory_mode)
	Sets the memory mode. More...
void	DisableLayerFusion ()
	Disables speedup due to layer fusion. More...
bool	Predict (float[] output_data, float[] input_data)
	Perform inference and obtain inference results. More...
Ailia.AILIAShape	GetInputShape ()
uint []	GetInputShapeND ()
	Obtains the shape of the input data at the time of inference. More...
bool	SetInputShape (Ailia.AILIAShape shape)
	Sets the shape of the input data during inference. More...
bool	SetInputShapeND (uint[] shape, int dim)
	Change the shape of the input data during inference. More...
Ailia.AILIAShape	GetOutputShape ()
	Obtains the shape of the output data during inference. More...
uint []	GetOutputShapeND ()
	Obtains the shape of the output data during inference. More...
Ailia.AILIAShape	GetBlobShape (string layer_name)
	Obtains the shape (layer format) of the internal data (Blob) at the time of inference. (Obsolete) More...
int	FindBlobIndexByName (string name)
	Look up and retrieve the index of the internal data (Blob) at the time of inference by name. More...
Ailia.AILIAShape	GetBlobShape (uint idx)
	Obtains the shape (layer format) of the internal data (Blob) at the time of inference. More...
Ailia.AILIAShape	GetBlobShape (int idx)
bool	GetBlobData (float[] output_data, uint idx)
	Obtains internal data (Blob) at the time of inference. More...
bool	GetBlobData (float[] output_data, int idx)
bool	SetInputBlobData (float[] input_data, uint idx)
	Set input data to the specified Blob. More...
bool	SetInputBlobData (float[] input_data, int idx)
bool	SetInputBlobShape (Ailia.AILIAShape shape, uint idx)
	Sets the format of the specified input data (Blob). (4 dimensions or less) More...
bool	SetInputBlobShape (Ailia.AILIAShape shape, int idx)
bool	SetInputBlobShapeND (uint[] shape, uint dim, uint idx)
	Sets the format of the input data (Blob). (5D or more) More...
bool	SetInputBlobShapeND (uint[] shape, int dim, int idx)
uint []	GetInputBlobList ()
	Obtains a list of the indices of the input data (Blob). More...
uint []	GetOutputBlobList ()
	Obtains a list of indices of the output data (Blob). More...
bool	CopyBlobData (uint dst_blob_idx, uint src_blob_idx, AiliaModel src_model=null)
	Perform copies between blobs that specified. More...
bool	Update ()
	Inference is performed with pre-populated data. More...
string	GetStatusString (int status)
	Returns a string corresponding to the status code. More...
string	GetErrorDetail ()
	Returns error details. More...
bool	SetProfileMode (uint profile_mode)
	Enable profile mode. More...
string	GetSummary ()
	Obtain network information and profile results. More...

Protected Member Functions
override void	Dispose (bool disposing)

Additional Inherited Members
Protected Attributes inherited from AiliaModel
IntPtr	ailia = IntPtr.Zero
bool	logging = true
Properties inherited from AiliaModel
int	Status [get, protected set]
	Status code Get the library status code. More...

Member Function Documentation

Close()

override void AiliaFeatureExtractorModel.Close ( )

inlinevirtual

Destroy feature extraction quobjects.

Returns

Return nothing.

Destroys the feature extractor object.

Reimplemented from AiliaModel.

                                {
        if(ailia_feature_extractor!=IntPtr.Zero){
            AiliaFeatureExtractor.ailiaDestroyFeatureExtractor(ailia_feature_extractor);
            ailia_feature_extractor=IntPtr.Zero;
        }
        base.Close();
    }

ComputeFromImage()

float [] AiliaFeatureExtractorModel.ComputeFromImage ( Color32 []  image, int  image_width, int  image_height  )

inline

Acquire features from images.

Parameters

image	Image to be extraction
image_width	Image width
image_height	Image height

Returns

Feature vector

Calculate feature vectors from images.

                                                                                      {
        return ComputeFromImageWithFormat(image,image_width,image_height,AiliaFormat.AILIA_IMAGE_FORMAT_RGBA);
    }

ComputeFromImageB2T()

float [] AiliaFeatureExtractorModel.ComputeFromImageB2T ( Color32 []  image, int  image_width, int  image_height  )

inline

Obtain features from the upside-down image.

Parameters

image	Image to be extraction
image_width	Image width
image_height	Image height

Returns

Feature vector

Calculate feature vectors from images.

                                                                                         {
        return ComputeFromImageWithFormat(image,image_width,image_height,AiliaFormat.AILIA_IMAGE_FORMAT_RGBA_B2T);
    }

Dispose() [1/2]

override void AiliaFeatureExtractorModel.Dispose ( )

inlinevirtual

Release resources.

Reimplemented from AiliaModel.

    {
        Dispose(true);
    }

Dispose() [2/2]

override void AiliaFeatureExtractorModel.Dispose ( bool  disposing )

inlineprotectedvirtual

Reimplemented from AiliaModel.

    {
        if (disposing){
            // release managed resource
        }
        Close(); // release unmanaged resource
        base.Dispose(disposing);
    }

Match()

float AiliaFeatureExtractorModel.Match ( float []  feature1, float []  feature2  )

inline

Calculates the distance between features.

Parameters

feature1	Feature 1
feature2	Feature 2

Returns

Calculated Distance

Calculate the distance between feature 1 and feature 2.

                                                           {
        if(feature1==null || feature2==null){
            if(logging){
                Debug.Log("input feature is empty");
            }
            return float.NaN;
        }

        float distance=0;

        GCHandle feature1_handle = GCHandle.Alloc(feature1, GCHandleType.Pinned);
        IntPtr feature1_buf_ptr = feature1_handle.AddrOfPinnedObject();

        GCHandle feature2_handle = GCHandle.Alloc(feature2, GCHandleType.Pinned);
        IntPtr feature2_buf_ptr = feature2_handle.AddrOfPinnedObject();

        int status=AiliaFeatureExtractor.ailiaFeatureExtractorMatch(ailia_feature_extractor,ref distance, distace_type, feature1_buf_ptr, (uint)feature1.Length*4, feature2_buf_ptr, (uint)feature2.Length*4);
        if(status!=Ailia.AILIA_STATUS_SUCCESS){
            if(logging){
                Debug.Log("ailiaFeatureExtractorMatch failed "+status);
            }
            return float.NaN;
        }

        feature1_handle.Free();
        feature2_handle.Free();
        return distance;
    }

OpenEx()

override bool AiliaFeatureExtractorModel.OpenEx ( Ailia.ailiaFileCallback  callback, IntPtr  arg1, IntPtr  arg2  )

inlinevirtual

Creates a network object from a file callback.

Parameters

callback	User-defined file access callback function structure
alg1	Argument pointer notified to AILIA_USER_API_FOPEN
alg2	Argument pointer notified to AILIA_USER_API_FOPEN

Returns

If this function is successful, it returns true , or false otherwise.

Creates a network object from a file callback.

Reimplemented from AiliaModel.

                                                                                         {
        Close();
        bool status=base.OpenEx(callback,arg1,arg2);
        if(status==false){
            if(logging){
                Debug.Log("ailiaModelOpenEx failed");
            }
            return false;
        }
        return OpenFeatureExtractor();
    }

OpenFile()

override bool AiliaFeatureExtractorModel.OpenFile ( string  prototxt, string  model_path  )

inlinevirtual

Create a network object from a model file.

Parameters

prototxt	Path name of the prototxt file (MBSC or UTF16)
model_path	Pathname of the protobuf/onnx file (MBSC or UTF16)

Returns

If this function is successful, it returns true , or false otherwise.

Create a network object from a model file.

Reimplemented from AiliaModel.

                                                                    {
        Close();
        bool status=base.OpenFile(prototxt,model_path);
        if(status==false){
            if(logging){
                Debug.Log("ailiaModelOpenFile failed");
            }
            return false;
        }
        return OpenFeatureExtractor();
    }

OpenMem()

override bool AiliaFeatureExtractorModel.OpenMem ( byte []  prototxt_buf, byte []  model_buf  )

inlinevirtual

Creates network objects from memory.

Parameters

prototxt	Pointer to data in prototxt file
model_path	Pointer to data in protobuf/onnx file

Returns

If this function is successful, it returns true , or false otherwise.

Creates network objects from memory.

Reimplemented from AiliaModel.

                                                                      {
        Close();
        bool status=base.OpenMem(prototxt_buf,model_buf);
        if(status==false){
            if(logging){
                Debug.Log("ailiaModelOpenMem failed");
            }
            return false;
        }
        return OpenFeatureExtractor();
    }

Settings()

bool AiliaFeatureExtractorModel.Settings ( uint  set_format, uint  set_channel, uint  set_range, uint  set_distance_type, string  set_layer_name  )

inline

Model setting.

Parameters

set_format	The network image format (AILIA_NETWORK_IMAGE_FORMAT_*)
set_channel	The network image channel (AILIA_NETWORK_IMAGE_CHANNEL_*)
set_range	The network image range (AILIA_NETWORK_IMAGE_RANGE_*)
set_distance_type	The type of the distance in feature space
set_layer_name	The name of the layer corresponding to the feature (fc1 for VGG16 and NULL for the last layer)

Returns

Returns true on success.

Pre-process images of the network and set up distance calculations.

                                                                                                                      {
        format=set_format;
        channel=set_channel;
        range=set_range;
        distace_type=set_distance_type;
        layer_name=set_layer_name;
        return true;
    }

---

The documentation for this class was generated from the following file:

• AiliaFeatureExtractorModel.cs