org.bytedeco.caffe

Class DoubleNet

java.lang.Object

org.bytedeco.javacpp.Pointer

org.bytedeco.caffe.DoubleNet

All Implemented Interfaces:

AutoCloseable

@Name(value="caffe::Net<double>")
 @NoOffset
 @Properties(inherit=caffe.class)
public class DoubleNet
extends Pointer

Nested Class Summary

Nested Classes

Modifier and Type	Class and Description
static class	DoubleNet.Callback

Nested classes/interfaces inherited from class org.bytedeco.javacpp.Pointer

Pointer.CustomDeallocator, Pointer.Deallocator, Pointer.NativeDeallocator

Field Summary

Fields inherited from class org.bytedeco.javacpp.Pointer

address, capacity, limit, position

Constructor Summary

Constructors

Constructor and Description
DoubleNet(BytePointer param_file, int phase)
DoubleNet(BytePointer param_file, int phase, int level, StringVector stages)
DoubleNet(NetParameter param)
DoubleNet(Pointer p) Pointer cast constructor.
DoubleNet(String param_file, int phase)
DoubleNet(String param_file, int phase, int level, StringVector stages)

Method Summary

Modifier and Type	Method and Description
void	add_after_backward(DoubleNet.Callback value)
void	add_after_forward(DoubleNet.Callback value)
void	add_before_backward(DoubleNet.Callback value)
void	add_before_forward(DoubleNet.Callback value)
PointerPointer	after_backward()
PointerPointer	after_forward()
void	Backward() The network backward should take no input and output, since it solely computes the gradient w.r.t the parameters, and the data has already been provided during the forward pass.
void	BackwardFrom(int start)
void	BackwardFromTo(int start, int end)
void	BackwardTo(int end)
PointerPointer	before_backward()
PointerPointer	before_forward()
DoubleBlob	blob_by_name(BytePointer blob_name)
DoubleBlob	blob_by_name(String blob_name)
DoublePointer	blob_loss_weights()
StringVector	blob_names() \brief returns the blob names
DoubleBlobSharedVector	blobs() \brief returns the blobs
IntPointer	bottom_ids(int i) \brief returns the ids of the bottom blobs of layer i
BoolVectorVector	bottom_need_backward()
DoubleBlobVectorVector	bottom_vecs() \brief returns the bottom vecs for each layer -- usually you won't need this unless you do per-layer checks such as gradients.
void	ClearParamDiffs() \brief Zeroes out the diffs of all net parameters.
void	CopyTrainedLayersFrom(BytePointer trained_filename)
void	CopyTrainedLayersFrom(NetParameter param) \brief For an already initialized net, copies the pre-trained layers from another Net.
void	CopyTrainedLayersFrom(String trained_filename)
void	CopyTrainedLayersFromBinaryProto(BytePointer trained_filename)
void	CopyTrainedLayersFromBinaryProto(String trained_filename)
void	CopyTrainedLayersFromHDF5(BytePointer trained_filename)
void	CopyTrainedLayersFromHDF5(String trained_filename)
static void	FilterNet(NetParameter param, NetParameter param_filtered) \brief Remove layers that the user specified should be excluded given the current phase, level, and stage.
DoubleBlobVector	Forward()
DoubleBlobVector	Forward(double[] loss)
DoubleBlobVector	Forward(DoubleBlobVector bottom)
DoubleBlobVector	Forward(DoubleBlobVector bottom, double[] loss)
DoubleBlobVector	Forward(DoubleBlobVector bottom, DoubleBuffer loss)
DoubleBlobVector	Forward(DoubleBlobVector bottom, DoublePointer loss) \brief DEPRECATED; set input blobs then use Forward() instead.
DoubleBlobVector	Forward(DoubleBuffer loss)
DoubleBlobVector	Forward(DoublePointer loss) \brief Run Forward and return the result.
double	ForwardBackward()
double	ForwardFrom(int start)
double	ForwardFromTo(int start, int end) The From and To variants of Forward and Backward operate on the (topological) ordering by which the net is specified.
DoubleBlobVector	ForwardPrefilled()
DoubleBlobVector	ForwardPrefilled(double[] loss)
DoubleBlobVector	ForwardPrefilled(DoubleBuffer loss)
DoubleBlobVector	ForwardPrefilled(DoublePointer loss) \brief DEPRECATED; use Forward() instead.
double	ForwardTo(int end)
boolean	has_blob(BytePointer blob_name)
boolean	has_blob(String blob_name)
boolean	has_layer(BytePointer layer_name)
boolean	has_layer(String layer_name)
BoolVector	has_params_decay()
BoolVector	has_params_lr()
void	Init(NetParameter param) \brief Initialize a network with a NetParameter.
IntPointer	input_blob_indices()
DoubleBlobVector	input_blobs()
DoubleLayer	layer_by_name(BytePointer layer_name)
<L extends DoubleLayer> L	layer_by_name(Class<L> cls, BytePointer layer_name)
<L extends DoubleLayer> L	layer_by_name(Class<L> cls, String layer_name)
DoubleLayer	layer_by_name(String layer_name)
StringVector	layer_names() \brief returns the layer names
BoolVector	layer_need_backward()
DoubleLayerSharedVector	layers() \brief returns the layers
DoubleBlobVector	learnable_params()
BytePointer	name() \brief returns the network name.
int	num_inputs() \brief Input and output blob numbers
int	num_outputs()
IntPointer	output_blob_indices()
DoubleBlobVector	output_blobs()
StringVector	param_display_names()
StringIntMap	param_names_index()
IntPointer	param_owners()
FloatPointer	params_lr() \brief returns the learnable parameter learning rate multipliers
FloatPointer	params_weight_decay() \brief returns the learnable parameter decay multipliers
DoubleBlobSharedVector	params() \brief returns the parameters
int	phase() \brief returns the phase: TRAIN or TEST
void	Reshape() \brief Reshape all layers from bottom to top.
void	set_debug_info(boolean value)
void	ShareTrainedLayersWith(DoubleNet other) \brief For an already initialized net, implicitly copies (i.e., using no additional memory) the pre-trained layers from another Net.
void	ShareWeights() \brief Shares weight data of owner blobs with shared blobs.
static boolean	StateMeetsRule(NetState state, NetStateRule rule, BytePointer layer_name) \brief return whether NetState state meets NetStateRule rule
static boolean	StateMeetsRule(NetState state, NetStateRule rule, String layer_name)
void	ToHDF5(BytePointer filename)
void	ToHDF5(BytePointer filename, boolean write_diff) \brief Writes the net to an HDF5 file.
void	ToHDF5(String filename)
void	ToHDF5(String filename, boolean write_diff)
IntPointer	top_ids(int i) \brief returns the ids of the top blobs of layer i
DoubleBlobVectorVector	top_vecs() \brief returns the top vecs for each layer -- usually you won't need this unless you do per-layer checks such as gradients.
void	ToProto(NetParameter param)
void	ToProto(NetParameter param, boolean write_diff) \brief Writes the net to a proto.
void	Update() \brief Updates the network weights based on the diff values computed.

Methods inherited from class org.bytedeco.javacpp.Pointer

address, asBuffer, asByteBuffer, availablePhysicalBytes, calloc, capacity, capacity, close, deallocate, deallocate, deallocateReferences, deallocator, deallocator, equals, fill, formatBytes, free, hashCode, isNull, limit, limit, malloc, maxBytes, maxPhysicalBytes, memchr, memcmp, memcpy, memmove, memset, offsetof, parseBytes, physicalBytes, position, position, put, realloc, setNull, sizeof, toString, totalBytes, totalPhysicalBytes, withDeallocator, zero

Methods inherited from class java.lang.Object

clone, finalize, getClass, notify, notifyAll, wait, wait, wait

Constructor Detail

DoubleNet

public DoubleNet(Pointer p)

Pointer cast constructor. Invokes Pointer.Pointer(Pointer).

DoubleNet

public DoubleNet(@Const @ByRef
                 NetParameter param)

DoubleNet

public DoubleNet(@StdString
                 BytePointer param_file,
                 @Cast(value="caffe::Phase")
                 int phase,
                 int level,
                 @Const
                 StringVector stages)

DoubleNet

public DoubleNet(@StdString
                 BytePointer param_file,
                 @Cast(value="caffe::Phase")
                 int phase)

DoubleNet

public DoubleNet(@StdString
                 String param_file,
                 @Cast(value="caffe::Phase")
                 int phase,
                 int level,
                 @Const
                 StringVector stages)

DoubleNet

public DoubleNet(@StdString
                 String param_file,
                 @Cast(value="caffe::Phase")
                 int phase)

Method Detail

Init

public void Init(@Const @ByRef
                 NetParameter param)

\brief Initialize a network with a NetParameter.

Forward

@Const
 @ByRef
public DoubleBlobVector Forward(DoublePointer loss)

\brief Run Forward and return the result.

Forward

@Const
 @ByRef
public DoubleBlobVector Forward()

Forward

@Const
 @ByRef
public DoubleBlobVector Forward(DoubleBuffer loss)

Forward

@Const
 @ByRef
public DoubleBlobVector Forward(double[] loss)

ForwardPrefilled

@Const
 @ByRef
public DoubleBlobVector ForwardPrefilled(DoublePointer loss)

\brief DEPRECATED; use Forward() instead.

ForwardPrefilled

@Const
 @ByRef
public DoubleBlobVector ForwardPrefilled()

ForwardPrefilled

@Const
 @ByRef
public DoubleBlobVector ForwardPrefilled(DoubleBuffer loss)

ForwardPrefilled

@Const
 @ByRef
public DoubleBlobVector ForwardPrefilled(double[] loss)

ForwardFromTo

public double ForwardFromTo(int start,
                            int end)

The From and To variants of Forward and Backward operate on the (topological) ordering by which the net is specified. For general DAG networks, note that (1) computing from one layer to another might entail extra computation on unrelated branches, and (2) computation starting in the middle may be incorrect if all of the layers of a fan-in are not included.

ForwardFrom

public double ForwardFrom(int start)

ForwardTo

public double ForwardTo(int end)

Forward

@Const
 @ByRef
public DoubleBlobVector Forward(@Const @ByRef
                                               DoubleBlobVector bottom,
                                               DoublePointer loss)

\brief DEPRECATED; set input blobs then use Forward() instead.

Forward

@Const
 @ByRef
public DoubleBlobVector Forward(@Const @ByRef
                                               DoubleBlobVector bottom)

Forward

@Const
 @ByRef
public DoubleBlobVector Forward(@Const @ByRef
                                               DoubleBlobVector bottom,
                                               DoubleBuffer loss)

Forward

@Const
 @ByRef
public DoubleBlobVector Forward(@Const @ByRef
                                               DoubleBlobVector bottom,
                                               double[] loss)

ClearParamDiffs

public void ClearParamDiffs()

\brief Zeroes out the diffs of all net parameters. Should be run before Backward.

Backward

public void Backward()

The network backward should take no input and output, since it solely computes the gradient w.r.t the parameters, and the data has already been provided during the forward pass.

BackwardFromTo

public void BackwardFromTo(int start,
                           int end)

BackwardFrom

public void BackwardFrom(int start)

BackwardTo

public void BackwardTo(int end)

Reshape

public void Reshape()

\brief Reshape all layers from bottom to top. This is useful to propagate changes to layer sizes without running a forward pass, e.g. to compute output feature size.

ForwardBackward

public double ForwardBackward()

Update

public void Update()

\brief Updates the network weights based on the diff values computed.

ShareWeights

public void ShareWeights()

\brief Shares weight data of owner blobs with shared blobs. Note: this is called by Net::Init, and thus should normally not be called manually.

ShareTrainedLayersWith

public void ShareTrainedLayersWith(@Const
                                   DoubleNet other)

\brief For an already initialized net, implicitly copies (i.e., using no additional memory) the pre-trained layers from another Net.

CopyTrainedLayersFrom

public void CopyTrainedLayersFrom(@Const @ByRef
                                  NetParameter param)

\brief For an already initialized net, copies the pre-trained layers from another Net.

CopyTrainedLayersFrom

public void CopyTrainedLayersFrom(@StdString
                                  BytePointer trained_filename)

CopyTrainedLayersFrom

public void CopyTrainedLayersFrom(@StdString
                                  String trained_filename)

CopyTrainedLayersFromBinaryProto

public void CopyTrainedLayersFromBinaryProto(@StdString
                                             BytePointer trained_filename)

CopyTrainedLayersFromBinaryProto

public void CopyTrainedLayersFromBinaryProto(@StdString
                                             String trained_filename)

CopyTrainedLayersFromHDF5

public void CopyTrainedLayersFromHDF5(@StdString
                                      BytePointer trained_filename)

CopyTrainedLayersFromHDF5

public void CopyTrainedLayersFromHDF5(@StdString
                                      String trained_filename)

ToProto

public void ToProto(NetParameter param,
                    @Cast(value="bool")
                    boolean write_diff)

\brief Writes the net to a proto.

ToProto

public void ToProto(NetParameter param)

ToHDF5

public void ToHDF5(@StdString
                   BytePointer filename,
                   @Cast(value="bool")
                   boolean write_diff)

\brief Writes the net to an HDF5 file.

ToHDF5

public void ToHDF5(@StdString
                   BytePointer filename)

ToHDF5

public void ToHDF5(@StdString
                   String filename,
                   @Cast(value="bool")
                   boolean write_diff)

ToHDF5

public void ToHDF5(@StdString
                   String filename)

name

@StdString
public BytePointer name()

\brief returns the network name.

layer_names

@Const
 @ByRef
public StringVector layer_names()

\brief returns the layer names

blob_names

@Const
 @ByRef
public StringVector blob_names()

\brief returns the blob names

blobs

@Const
 @ByRef
public DoubleBlobSharedVector blobs()

\brief returns the blobs

layers

@Const
 @ByRef
public DoubleLayerSharedVector layers()

\brief returns the layers

phase

@Cast(value="caffe::Phase")
public int phase()

\brief returns the phase: TRAIN or TEST

bottom_vecs

@Const
 @ByRef
public DoubleBlobVectorVector bottom_vecs()

\brief returns the bottom vecs for each layer -- usually you won't need this unless you do per-layer checks such as gradients.

top_vecs

@Const
 @ByRef
public DoubleBlobVectorVector top_vecs()

\brief returns the top vecs for each layer -- usually you won't need this unless you do per-layer checks such as gradients.

top_ids

@StdVector
public IntPointer top_ids(int i)

\brief returns the ids of the top blobs of layer i

bottom_ids

@StdVector
public IntPointer bottom_ids(int i)

\brief returns the ids of the bottom blobs of layer i

bottom_need_backward

@Const
 @ByRef
public BoolVectorVector bottom_need_backward()

blob_loss_weights

@StdVector
public DoublePointer blob_loss_weights()

layer_need_backward

@Const
 @ByRef
public BoolVector layer_need_backward()

params

@Const
 @ByRef
public DoubleBlobSharedVector params()

\brief returns the parameters

learnable_params

@Const
 @ByRef
public DoubleBlobVector learnable_params()

params_lr

@StdVector
public FloatPointer params_lr()

\brief returns the learnable parameter learning rate multipliers

has_params_lr

@Const
 @ByRef
public BoolVector has_params_lr()

params_weight_decay

@StdVector
public FloatPointer params_weight_decay()

\brief returns the learnable parameter decay multipliers

has_params_decay

@Const
 @ByRef
public BoolVector has_params_decay()

param_names_index

@Const
 @ByRef
public StringIntMap param_names_index()

param_owners

@StdVector
public IntPointer param_owners()

param_display_names

@Const
 @ByRef
public StringVector param_display_names()

num_inputs

public int num_inputs()

\brief Input and output blob numbers

num_outputs

public int num_outputs()

input_blobs

@Const
 @ByRef
public DoubleBlobVector input_blobs()

output_blobs

@Const
 @ByRef
public DoubleBlobVector output_blobs()

input_blob_indices

@StdVector
public IntPointer input_blob_indices()

output_blob_indices

@StdVector
public IntPointer output_blob_indices()

has_blob

@Cast(value="bool")
public boolean has_blob(@StdString
                                            BytePointer blob_name)

has_blob

@Cast(value="bool")
public boolean has_blob(@StdString
                                            String blob_name)

blob_by_name

@Const
 @SharedPtr
 @ByVal
public DoubleBlob blob_by_name(@StdString
                                                          BytePointer blob_name)

blob_by_name

@Const
 @SharedPtr
 @ByVal
public DoubleBlob blob_by_name(@StdString
                                                          String blob_name)

has_layer

@Cast(value="bool")
public boolean has_layer(@StdString
                                             BytePointer layer_name)

has_layer

@Cast(value="bool")
public boolean has_layer(@StdString
                                             String layer_name)

layer_by_name

public DoubleLayer layer_by_name(BytePointer layer_name)

layer_by_name

public DoubleLayer layer_by_name(String layer_name)

layer_by_name

@Const
 @Cast(value={"","boost::shared_ptr<caffe::Layer<double> >"})
 @SharedPtr
 @ByVal
public <L extends DoubleLayer> L layer_by_name(Class<L> cls,
                                                                                                                                        @StdString
                                                                                                                                        BytePointer layer_name)

layer_by_name

@Const
 @Cast(value={"","boost::shared_ptr<caffe::Layer<double> >"})
 @SharedPtr
 @ByVal
public <L extends DoubleLayer> L layer_by_name(Class<L> cls,
                                                                                                                                        @StdString
                                                                                                                                        String layer_name)

set_debug_info

public void set_debug_info(@Cast(value="const bool")
                           boolean value)

FilterNet

public static void FilterNet(@Const @ByRef
                             NetParameter param,
                             NetParameter param_filtered)

\brief Remove layers that the user specified should be excluded given the current phase, level, and stage.

StateMeetsRule

@Cast(value="bool")
public static boolean StateMeetsRule(@Const @ByRef
                                                         NetState state,
                                                         @Const @ByRef
                                                         NetStateRule rule,
                                                         @StdString
                                                         BytePointer layer_name)

\brief return whether NetState state meets NetStateRule rule

StateMeetsRule

@Cast(value="bool")
public static boolean StateMeetsRule(@Const @ByRef
                                                         NetState state,
                                                         @Const @ByRef
                                                         NetStateRule rule,
                                                         @StdString
                                                         String layer_name)

before_forward

@Cast(value="caffe::Net<double>::Callback**")
 @StdVector
public PointerPointer before_forward()

add_before_forward

public void add_before_forward(DoubleNet.Callback value)

after_forward

@Cast(value="caffe::Net<double>::Callback**")
 @StdVector
public PointerPointer after_forward()

add_after_forward

public void add_after_forward(DoubleNet.Callback value)

before_backward

@Cast(value="caffe::Net<double>::Callback**")
 @StdVector
public PointerPointer before_backward()

add_before_backward

public void add_before_backward(DoubleNet.Callback value)

after_backward

@Cast(value="caffe::Net<double>::Callback**")
 @StdVector
public PointerPointer after_backward()

add_after_backward

public void add_after_backward(DoubleNet.Callback value)