#include <ImFusion/ML/OperationsSequence.h>

Class to execute a sequence of Operation, depending on the workflow phase (training, validation, etc.). More...

Detailed Description

Class to execute a sequence of Operation, depending on the workflow phase (training, validation, etc.).

Public Types
using	PhasedOperation = std::pair<std::shared_ptr<Operation>, Phase>

Public Member Functions
	OperationsSequence (bool verbose=false)

	OperationsSequence (const std::vector< Operation::Specs > &pipelineConfig, bool verbose=false)
	Constructs an operation sequence from a sequence of specification, each for a different operation.

	OperationsSequence (const OperationsSequence &that)=delete

OperationsSequence &	operator= (const OperationsSequence &that)=delete

	OperationsSequence (OperationsSequence &&that)=default

OperationsSequence &	operator= (OperationsSequence &&that)=default

Status	addOperation (const Operation::Specs &specs)
	Add an operation specified by a configuration spec.

Status	addOperation (std::shared_ptr< Operation > op, Phase phase=Phase::Always)
	Add an existing operation.

Status	process (DataItem &item, Phase phase=Phase::Always) const
	Apply sequence of operations on all elements of the data item.

std::shared_ptr< SharedImageSet >	process (std::shared_ptr< SharedImageSet > input, Phase phase=Phase::Always) const
	Helper function to apply the sequence of operations on one input image.

std::pair< std::shared_ptr< SharedImageSet >, std::shared_ptr< SharedImageSet > >	process (std::shared_ptr< SharedImageSet > input, std::shared_ptr< SharedImageSet > label, Phase phase=Phase::Always) const
	Helper function to apply the sequence of operations on input image and label.

std::vector< std::string >	getOperationNames () const
	Returns the names of the operations added so far.

std::vector< std::shared_ptr< Operation > >	operations ()
	Returns a vector of shared_ptrs to the operations added so far.

bool	ok () const
	Returns whether the operations sequence setup was successful.

Status	status () const
	Returns status of the operations sequence setup.

void	resetOperationSeeds ()
	Resets the random generator seeding of all the operations that where explicitly configured with a "seed" parameter.

void	setOperationSeeds (unsigned int seed)
	Sets the random generator seeding of all the operations.

void	setVerbose (bool verbose)
	Set the verbose flag.

void	setOperationsComputingDevice (ML::ComputingDevice device)
	Set the computing device of all current operations.

void	setErrorOnUnexpectedBehaviour (bool treatWarningsAsErrors)
	Treat unexpected behaviour warnings as errors.

bool	doesNotModifyInput () const
	Returns true if the operation is guaranteed to not modify its input (either because it computes its output out-of-place or because it is a no-op).

Constructor & Destructor Documentation

◆ OperationsSequence()

OperationsSequence	(	const std::vector< Operation::Specs > &	pipelineConfig,
		bool	verbose = false )

explicit

Constructs an operation sequence from a sequence of specification, each for a different operation.

At each call to process each operation is applied in the order they were specified.

Parameters

pipelineConfig processing pipeline configuration sequence

Member Function Documentation

◆ addOperation() [1/2]

Status addOperation ( const Operation::Specs & specs )

Add an operation specified by a configuration spec.

Note: The required operation is created by the OperationFactory

◆ addOperation() [2/2]

Status addOperation	(	std::shared_ptr< Operation >	op,
		Phase	phase = Phase::Always )

Add an existing operation.

Parameters

op	the operation to add
phase	limits application of the op to a specific phase (i.e. Training, or Test)

◆ doesNotModifyInput()

bool doesNotModifyInput ( ) const

Returns true if the operation is guaranteed to not modify its input (either because it computes its output out-of-place or because it is a no-op).

This can be used to decide whether a copy of the input should be made before calling the OperationsSequence::process function. Note that this only relates to the overloads taking specific elements as inputs (e.g. images, keypoints, etc.), since the OperationsSequence::process(DataItem& item) function always modifies its input.

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

ImFusion/ML/OperationsSequence.h