Public Member Functions | Static Public Attributes | Protected Attributes
ngraph::op::v1::DeformableConvolution Class Reference

DeformableConvolution operation. More...

#include <deformable_convolution.hpp>

Inheritance diagram for ngraph::op::v1::DeformableConvolution:
Inheritance graph
Collaboration diagram for ngraph::op::v1::DeformableConvolution:
Collaboration graph

Public Member Functions

const NodeTypeInfoget_type_info () const override
 DeformableConvolution ()=default
 Constructs a conversion operation.
 DeformableConvolution (const Output< Node > &arg, const Output< Node > &deformable_values, const Output< Node > &filters, const Strides &strides, const CoordinateDiff &pads_begin, const CoordinateDiff &pads_end, const Strides &dilations, const PadType &auto_pad=PadType::EXPLICIT, const int64_t group=1, const int64_t deformable_group=1)
 Constructs a conversion operation. More...
bool visit_attributes (AttributeVisitor &visitor) override
void validate_and_infer_types () override
 Verifies that attributes and inputs are consistent and computes output shapes and element types. Must be implemented by concrete child classes so that it can be run any number of times. More...
const Stridesget_strides () const
void set_strides (const Strides &strides)
const Stridesget_dilations () const
void set_dilations (const Strides &dilations)
const CoordinateDiffget_pads_begin () const
void set_pads_begin (const CoordinateDiff &pads_begin)
const CoordinateDiffget_pads_end () const
void set_pads_end (const CoordinateDiff &pads_end)
const PadTypeget_auto_pad () const
void set_auto_pad (const PadType &auto_pad)
int64_t get_group () const
void set_group (const int64_t group)
int64_t get_deformable_group () const
void set_deformable_group (const int64_t deformable_group)
virtual std::shared_ptr< Nodeclone_with_new_inputs (const OutputVector &new_args) const override
- Public Member Functions inherited from ngraph::Node
void constructor_validate_and_infer_types ()
virtual const op::AutoBroadcastSpecget_autob () const
virtual bool evaluate (const HostTensorVector &output_values, const HostTensorVector &input_values) const
 Evaluates the op on input_values putting results in output_values. More...
virtual bool evaluate_lower (const HostTensorVector &output_values) const
virtual bool evaluate_upper (const HostTensorVector &output_values) const
virtual bool constant_fold (OutputVector &output_values, const OutputVector &inputs_values)
virtual OutputVector decompose_op () const
 Decomposes the FusedOp into a sub-graph consisting of core ngraph ops. More...
const char * get_type_name () const
void set_arguments (const NodeVector &arguments)
 Sets/replaces the arguments with new arguments.
void set_arguments (const OutputVector &arguments)
 Sets/replaces the arguments with new arguments.
void set_argument (size_t position, const Output< Node > &argument)
 Sets/replaces the arguments with new arguments.
void set_output_type (size_t i, const element::Type &element_type, const PartialShape &pshape)
void set_output_size (size_t output_size)
 Sets the number of outputs.
void invalidate_values ()
virtual void revalidate_and_infer_types ()
virtual std::string description () const
 Get the string name for the type of the node, such as Add or Multiply. The class name, must not contain spaces as it is used for codegen. More...
const std::string & get_name () const
 Get the unique name of the node. More...
void set_friendly_name (const std::string &name)
 Sets a friendly name for a node. This does not overwrite the unique name of the node and is retrieved via get_friendly_name(). Used mainly for debugging. The friendly name may be set exactly once. More...
const std::string & get_friendly_name () const
 Gets the friendly name for a node. If no friendly name has been set via set_friendly_name then the node's unique name is returned. More...
virtual bool is_dynamic () const
size_t get_instance_id () const
virtual std::ostream & write_description (std::ostream &os, uint32_t depth=0) const
 Writes a description of a node to a stream. More...
const std::vector< std::shared_ptr< Node > > & get_control_dependencies () const
 Get control dependencies registered on the node.
const std::vector< Node * > & get_control_dependents () const
 Get nodes dependent on this node.
void add_control_dependency (std::shared_ptr< Node > node)
 This node cannot execute until node executes.
void remove_control_dependency (std::shared_ptr< Node > node)
 Remove the dependency of this node on node.
void clear_control_dependencies ()
 Remove all dependencies from this node.
void clear_control_dependents ()
 Remove this node as a dependency from all dependent nodes.
void add_node_control_dependencies (std::shared_ptr< Node > source_node)
 This node absorbs the control dependencies of source_node.
void add_node_control_dependents (std::shared_ptr< Node > source_node)
 This node becomes a dependent of every node dependent on source_node.
void transfer_control_dependents (std::shared_ptr< Node > replacement)
 This node's control dependencies are replaced by replacement.
size_t get_output_size () const
 Returns the number of outputs from the node.
const element::Typeget_output_element_type (size_t i) const
 Returns the element type for output i.
const element::Typeget_element_type () const
 Checks that there is exactly one output and returns its element type.
const Shapeget_output_shape (size_t i) const
 Returns the shape for output i.
const PartialShapeget_output_partial_shape (size_t i) const
 Returns the partial shape for output i.
Output< const Nodeget_default_output () const
Output< Nodeget_default_output ()
virtual size_t get_default_output_index () const
 Returns the output of the default output, or throws if there is none.
size_t no_default_index () const
 Throws no default.
const Shapeget_shape () const
 Checks that there is exactly one output and returns its shape.
descriptor::Tensorget_output_tensor (size_t i) const
 Returns the tensor for output or input i.
descriptor::Tensorget_input_tensor (size_t i) const
size_t get_input_size () const
 Returns the number of inputs for the op.
const element::Typeget_input_element_type (size_t i) const
 Returns the element type of input i.
const Shapeget_input_shape (size_t i) const
 Returns the shape of input i.
const PartialShapeget_input_partial_shape (size_t i) const
 Returns the partial shape of input i.
Nodeget_input_node_ptr (size_t index) const
std::shared_ptr< Nodeget_input_node_shared_ptr (size_t index) const
Output< Nodeget_input_source_output (size_t i) const
std::shared_ptr< Nodecopy_with_new_inputs (const OutputVector &new_args) const
std::shared_ptr< Nodecopy_with_new_inputs (const OutputVector &inputs, const std::vector< std::shared_ptr< Node >> &control_dependencies) const
bool has_same_type (std::shared_ptr< const Node > node) const
 True if this and node have one output with same element type and shape.
RTMap & get_rt_info ()
const RTMap & get_rt_info () const
const std::unordered_set< std::string > & get_provenance_tags () const
void add_provenance_tag (const std::string &tag)
template<typename T >
void add_provenance_tags (T tag_set)
void add_provenance_tags_above (const OutputVector &base, const std::unordered_set< std::string > &tag_set)
 Adds tag_set to this node and all intermediate nodes above base.
void remove_provenance_tag (const std::string &tag)
void add_provenance_group_member (const std::shared_ptr< Node > &node)
 Add node to additional nodes that receive tags.
void remove_provenance_group_member (const std::shared_ptr< Node > &node)
 Remove node to additional nodes that receive tags.
void replace_provenance_group_member (const std::shared_ptr< Node > &current_node, const std::shared_ptr< Node > &replacement_node)
 Replace current_node with replacement_node and transfer tags.
const std::set< std::shared_ptr< Node > > & get_provenance_group_members () const
std::shared_ptr< Nodeadd_provenance_group_members_above (const OutputVector &base)
 Add all nodes between this node and nodes in base as additional nodes to receive provenance tags.
void merge_provenance_tags_from (const std::shared_ptr< const Node > &source)
void transfer_provenance_tags (const std::shared_ptr< Node > &replacement)
 Transfer provenance tags to replacement.
NodeVector get_users (bool check_is_used=false) const
 Get all the nodes that uses the current node.
virtual size_t get_version () const
virtual std::shared_ptr< Nodeget_default_value () const
bool operator< (const Node &other) const
 Use instance ids for comparison instead of memory addresses to improve determinism.
std::vector< Input< Node > > inputs ()
std::vector< Input< const Node > > inputs () const
std::vector< Output< Node > > input_values () const
std::vector< Output< Node > > outputs ()
std::vector< Output< const Node > > outputs () const
Input< Nodeinput (size_t input_index)
Input< const Nodeinput (size_t input_index) const
Output< Nodeinput_value (size_t input_index) const
Output< Nodeoutput (size_t output_index)
Output< const Nodeoutput (size_t output_index) const
void set_op_annotations (std::shared_ptr< ngraph::op::util::OpAnnotations > op_annotations)
std::shared_ptr< ngraph::op::util::OpAnnotationsget_op_annotations () const
virtual bool match_value (pattern::Matcher *matcher, const Output< Node > &pattern_value, const Output< Node > &graph_value)
virtual bool match_node (pattern::Matcher *matcher, const Output< Node > &graph_value)

Static Public Attributes

static constexpr NodeTypeInfo type_info {"DeformableConvolution", 1}

Protected Attributes

Strides m_strides
Strides m_dilations
CoordinateDiff m_pads_begin
CoordinateDiff m_pads_end
PadType m_auto_pad
int64_t m_group
int64_t m_deformable_group

Additional Inherited Members

- Public Types inherited from ngraph::Node
using type_info_t = DiscreteTypeInfo
using RTMap = std::map< std::string, std::shared_ptr< Variant > >
- Data Fields inherited from ngraph::Node
NGRAPH_DEPRECATED("The tensor name was deprecated. Use get_input_tensor(i).get_names() instead.") const std std::unordered_set< descriptor::Tensor * > liveness_new_list
 Returns the tensor name for input i.
std::unordered_set< descriptor::Tensor * > liveness_free_list
- Protected Member Functions inherited from ngraph::op::Op
 Op (const OutputVector &arguments)
- Protected Member Functions inherited from ngraph::Node
 Node ()=default
 Construct an unitialized Node.
 Node (const Node &)
 Copying a node.
Nodeoperator= (const Node &)
 Assignment operator.
 Node (size_t output_size)
 Construct an unitialized Node. More...
 Node (const OutputVector &arguments, size_t output_size=1)
 Constructor for Node subclasses that have metaclasses. More...
void safe_delete (NodeVector &nodes, bool recurse)
 Moves nodes that would be deleted from inputs to nodes to avoid stack overflows on deep networks.
void set_input_is_relevant_to_shape (size_t i, bool relevant=true)
 Marks an input as being relevant or irrelevant to the output shapes of this node. More...
void set_input_is_relevant_to_value (size_t i, bool relevant=true)
 Marks an input as being relevant or irrelevant to the output values of this node. More...

Detailed Description

DeformableConvolution operation.

Constructor & Destructor Documentation

◆ DeformableConvolution()

ngraph::op::v1::DeformableConvolution::DeformableConvolution ( const Output< Node > &  arg,
const Output< Node > &  deformable_values,
const Output< Node > &  filters,
const Strides strides,
const CoordinateDiff pads_begin,
const CoordinateDiff pads_end,
const Strides dilations,
const PadType auto_pad = PadType::EXPLICIT,
const int64_t  group = 1,
const int64_t  deformable_group = 1 

Constructs a conversion operation.

argNode that produces the input tensor.
deformable_valuesNode producing the deformable values tensor.
filtersNode producing the filters(kernels) tensor wit OIZYX layout.
stridesConvolution strides.
pads_beginAmount of padding to be added to the beginning along each axis. For example in case of a 2D input the value of (1, 2) means that 1 element will be added to the top and 2 elements to the left.
pads_endAmount of padding to be added to the end along each axis.
dilationsThe distance in width and height between the weights in the filters tensor.
auto_padSpecifies how the automatic calculation of padding should be done.
groupThe number of groups which both output and input should be split into.
deformable_groupThe number of groups which deformable values and output should be split into along the channel axis.

Member Function Documentation

◆ get_type_info()

const NodeTypeInfo& ngraph::op::v1::DeformableConvolution::get_type_info ( ) const

Returns the NodeTypeInfo for the node's class. During transition to type_info, returns a dummy type_info for Node if the class has not been updated yet.

Implements ngraph::Node.

◆ validate_and_infer_types()

void ngraph::op::v1::DeformableConvolution::validate_and_infer_types ( )

Verifies that attributes and inputs are consistent and computes output shapes and element types. Must be implemented by concrete child classes so that it can be run any number of times.

Throws if the node is invalid.

Reimplemented from ngraph::Node.

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