LSTM recurrent layer. More...

#include <opencv2/dnn/all_layers.hpp>

Inheritance diagram for cv::dnn::LSTMLayer:

Public Member Functions
int	inputNameToIndex (String inputName) CV_OVERRIDE
	Returns index of input blob into the input array.

int	outputNameToIndex (const String &outputName) CV_OVERRIDE
	Returns index of output blob in output array.

virtual void	setOutShape (const MatShape &outTailShape=MatShape())=0
	Specifies shape of output blob which will be [[`T`], `N`] + `outTailShape`.

virtual void	setProduceCellOutput (bool produce=false)=0
	If this flag is set to true then layer will produce \( c_t \) as second output.

virtual void	setUseTimstampsDim (bool use=true)=0
	Specifies either interpret first dimension of input blob as timestamp dimenion either as sample.

virtual void	setWeights (const Mat &Wh, const Mat &Wx, const Mat &b)=0
	Set trained weights for LSTM layer.

Public Member Functions inherited from cv::dnn::Layer
	Layer ()

	Layer (const LayerParams &params)
	Initializes only name, type and blobs fields.

virtual	~Layer ()

virtual void	applyHalideScheduler (Ptr< BackendNode > &node, const std::vector< Mat * > &inputs, const std::vector< Mat > &outputs, int targetId) const
	Automatic Halide scheduling based on layer hyper-parameters.

virtual void	finalize (const std::vector< Mat * > &input, std::vector< Mat > &output)
	Computes and sets internal parameters according to inputs, outputs and blobs.

virtual void	finalize (InputArrayOfArrays inputs, OutputArrayOfArrays outputs)
	Computes and sets internal parameters according to inputs, outputs and blobs.

void	finalize (const std::vector< Mat > &inputs, std::vector< Mat > &outputs)
	This is an overloaded member function, provided for convenience. It differs from the above function only in what argument(s) it accepts.

std::vector< Mat >	finalize (const std::vector< Mat > &inputs)
	This is an overloaded member function, provided for convenience. It differs from the above function only in what argument(s) it accepts.

virtual void	forward (std::vector< Mat * > &input, std::vector< Mat > &output, std::vector< Mat > &internals)
	Given the `input` blobs, computes the output `blobs`.

virtual void	forward (InputArrayOfArrays inputs, OutputArrayOfArrays outputs, OutputArrayOfArrays internals)
	Given the `input` blobs, computes the output `blobs`.

void	forward_fallback (InputArrayOfArrays inputs, OutputArrayOfArrays outputs, OutputArrayOfArrays internals)
	Given the `input` blobs, computes the output `blobs`.

virtual int64	getFLOPS (const std::vector< MatShape > &inputs, const std::vector< MatShape > &outputs) const

virtual bool	getMemoryShapes (const std::vector< MatShape > &inputs, const int requiredOutputs, std::vector< MatShape > &outputs, std::vector< MatShape > &internals) const

virtual void	getScaleShift (Mat &scale, Mat &shift) const
	Returns parameters of layers with channel-wise multiplication and addition.

virtual Ptr< BackendNode >	initHalide (const std::vector< Ptr< BackendWrapper > > &inputs)
	Returns Halide backend node.

virtual Ptr< BackendNode >	initInfEngine (const std::vector< Ptr< BackendWrapper > > &inputs)

virtual Ptr< BackendNode >	initVkCom (const std::vector< Ptr< BackendWrapper > > &inputs)

void	run (const std::vector< Mat > &inputs, std::vector< Mat > &outputs, std::vector< Mat > &internals)
	Allocates layer and computes output.

virtual bool	setActivation (const Ptr< ActivationLayer > &layer)
	Tries to attach to the layer the subsequent activation layer, i.e. do the layer fusion in a partial case.

void	setParamsFrom (const LayerParams &params)
	Initializes only name, type and blobs fields.

virtual bool	supportBackend (int backendId)
	Ask layer if it support specific backend for doing computations.

virtual Ptr< BackendNode >	tryAttach (const Ptr< BackendNode > &node)
	Implement layers fusing.

virtual bool	tryFuse (Ptr< Layer > &top)
	Try to fuse current layer with a next one.

virtual void	unsetAttached ()
	"Deattaches" all the layers, attached to particular layer.

Public Member Functions inherited from cv::Algorithm
	Algorithm ()

virtual	~Algorithm ()

virtual void	clear ()
	Clears the algorithm state.

virtual bool	empty () const
	Returns true if the Algorithm is empty (e.g. in the very beginning or after unsuccessful read.

virtual String	getDefaultName () const

virtual void	read (const FileNode &fn)
	Reads algorithm parameters from a file storage.

virtual void	save (const String &filename) const

virtual void	write (FileStorage &fs) const
	Stores algorithm parameters in a file storage.

void	write (const Ptr< FileStorage > &fs, const String &name=String()) const
	simplified API for language bindingsThis is an overloaded member function, provided for convenience. It differs from the above function only in what argument(s) it accepts.

Static Public Member Functions
static Ptr< LSTMLayer >	create (const LayerParams &params)

Additional Inherited Members
Public Attributes inherited from cv::dnn::Layer
std::vector< Mat >	blobs
	List of learned parameters must be stored here to allow read them by using Net::getParam().

String	name
	Name of the layer instance, can be used for logging or other internal purposes.

int	preferableTarget
	prefer target for layer forwarding

String	type
	Type name which was used for creating layer by layer factory.

Protected Member Functions inherited from cv::Algorithm
void	writeFormat (FileStorage &fs) const

Detailed Description

LSTM recurrent layer.

Member Function Documentation

static Ptr<LSTMLayer> cv::dnn::LSTMLayer::create ( const LayerParams & params )

static

Creates instance of LSTM layer

int cv::dnn::LSTMLayer::inputNameToIndex ( String inputName )

virtual

Returns index of input blob into the input array.

Parameters

inputName label of input blob

Each layer input and output can be labeled to easily identify them using "%<layer_name%>[.output_name]" notation. This method maps label of input blob to its index into input vector.

Reimplemented from cv::dnn::Layer.

int cv::dnn::LSTMLayer::outputNameToIndex ( const String & outputName )

virtual

Returns index of output blob in output array.

See Also: inputNameToIndex()

Reimplemented from cv::dnn::Layer.

virtual void cv::dnn::LSTMLayer::setOutShape ( const MatShape & outTailShape = MatShape() )

pure virtual

Specifies shape of output blob which will be [[T], N] + outTailShape.

If this parameter is empty or unset then outTailShape = [Wh.size(0)] will be used, where Wh is parameter from setWeights().

virtual void cv::dnn::LSTMLayer::setProduceCellOutput ( bool produce = false )

pure virtual

If this flag is set to true then layer will produce \( c_t \) as second output.

Deprecated:: Use flag use_timestamp_dim in LayerParams.

Shape of the second output is the same as first output.

virtual void cv::dnn::LSTMLayer::setUseTimstampsDim ( bool use = true )

pure virtual

Specifies either interpret first dimension of input blob as timestamp dimenion either as sample.

Deprecated:: Use flag produce_cell_output in LayerParams.

If flag is set to true then shape of input blob will be interpreted as [T, N, [data dims]] where T specifies number of timestamps, N is number of independent streams. In this case each forward() call will iterate through T timestamps and update layer's state T times.

If flag is set to false then shape of input blob will be interpreted as [N, [data dims]]. In this case each forward() call will make one iteration and produce one timestamp with shape [N, [out dims]].

virtual void cv::dnn::LSTMLayer::setWeights	(	const Mat &	Wh,
		const Mat &	Wx,
		const Mat &	b
	)

pure virtual

Set trained weights for LSTM layer.

Deprecated:: Use LayerParams::blobs instead.

LSTM behavior on each step is defined by current input, previous output, previous cell state and learned weights.

Let \(x_t\) be current input, \(h_t\) be current output, \(c_t\) be current state. Than current output and current cell state is computed as follows:

\begin{eqnarray*} h_t &= o_t \odot tanh(c_t), \\ c_t &= f_t \odot c_{t-1} + i_t \odot g_t, \\ \end{eqnarray*}

where \(\odot\) is per-element multiply operation and \(i_t, f_t, o_t, g_t\) is internal gates that are computed using learned wights.

Gates are computed as follows:

\begin{eqnarray*} i_t &= sigmoid&(W_{xi} x_t + W_{hi} h_{t-1} + b_i), \\ f_t &= sigmoid&(W_{xf} x_t + W_{hf} h_{t-1} + b_f), \\ o_t &= sigmoid&(W_{xo} x_t + W_{ho} h_{t-1} + b_o), \\ g_t &= tanh &(W_{xg} x_t + W_{hg} h_{t-1} + b_g), \\ \end{eqnarray*}

where \(W_{x?}\), \(W_{h?}\) and \(b_{?}\) are learned weights represented as matrices: \(W_{x?} \in R^{N_h \times N_x}\), \(W_{h?} \in R^{N_h \times N_h}\), \(b_? \in R^{N_h}\).

For simplicity and performance purposes we use \( W_x = [W_{xi}; W_{xf}; W_{xo}, W_{xg}] \) (i.e. \(W_x\) is vertical concatenation of \( W_{x?} \)), \( W_x \in R^{4N_h \times N_x} \). The same for \( W_h = [W_{hi}; W_{hf}; W_{ho}, W_{hg}], W_h \in R^{4N_h \times N_h} \) and for \( b = [b_i; b_f, b_o, b_g]\), \(b \in R^{4N_h} \).

Parameters

Wh	is matrix defining how previous output is transformed to internal gates (i.e. according to above mentioned notation is \( W_h \))
Wx	is matrix defining how current input is transformed to internal gates (i.e. according to above mentioned notation is \( W_x \))
b	is bias vector (i.e. according to above mentioned notation is \( b \))

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

all_layers.hpp

Public Member Functions

Static Public Member Functions

Additional Inherited Members

Detailed Description

Member Function Documentation