cherrypi::UnitAttributeFeaturizer Struct Reference

Abstract base class for featurizing unit attributes in a sparse manner. More...

#include <unitsfeatures.h>

Inherited by cherrypi::UnitFlagsFeaturizer, cherrypi::UnitPresenceFeaturizer, cherrypi::UnitStatFeaturizer, cherrypi::UnitTypeDefoggerFeaturizer, and cherrypi::UnitTypeFeaturizer.

Classes
struct	Data

Public Types
using	UnitFilter = std::function< bool(Unit *)>
using	TensorDest = torch::TensorAccessor< float, 1 >

Public Member Functions
virtual	~UnitAttributeFeaturizer ()=default
virtual Data	extract (State *state, UnitsInfo::Units const &units, Rect const &boundingBox=Rect())
	Extract unit features for a given set of units. More...
Data	extract (State *state, Rect const &boundingBox=Rect())
	Extract unit features for all live units. More...
Data	extract (State *state, UnitFilter filter, Rect const &boundingBox=Rect())
	Extract unit features for all live units that pass the given filter. More...
FeatureData	toSpatialFeature (Data const &data, SubsampleMethod pooling=SubsampleMethod::Sum) const
	Embeds the unit attribute data into a spatial feature. More...
void	toSpatialFeature (FeatureData *dest, Data const &data, SubsampleMethod pooling=SubsampleMethod::Sum) const
	Embeds the unit attribute data into a spatial feature. More...

Public Attributes
std::shared_ptr< BaseJitter >	jitter = std::make_shared<NoJitter>()
	Optional jittering of unit positions. More...
CustomFeatureType	type
std::string	name
int	numChannels

Protected Member Functions
virtual void	extractUnit (TensorDest acc, Unit *unit)=0
	Reimplement this in actual featurizers. More...

Detailed Description

Abstract base class for featurizing unit attributes in a sparse manner.

General usage of sub-classes for actual feature extraction boils down to calling extract() with a desired subset of units to featurize. The resulting data is sparse wrt positions, i.e. it contains a tensor of positions for each unit and the accompanying data as defined by a featurizer implementation.

toSpatialFeature() will transform the given data to a FeatureData object, i.e. it will place the feature data at the respective positions.

Optionally, users can set a jittering method that will be accounted for in extract().

Member Typedef Documentation

using cherrypi::UnitAttributeFeaturizer::TensorDest = torch::TensorAccessor<float, 1>

using cherrypi::UnitAttributeFeaturizer::UnitFilter = std::function<bool(Unit*)>

Constructor & Destructor Documentation

virtual cherrypi::UnitAttributeFeaturizer::~UnitAttributeFeaturizer ( )

virtualdefault

Member Function Documentation

UnitAttributeFeaturizer::Data cherrypi::UnitAttributeFeaturizer::extract ( State *  state, UnitsInfo::Units const &  units, Rect const &  boundingBox = Rect()  )

virtual

Extract unit features for a given set of units.

Reimplemented in cherrypi::UnitTypeMDefoggerFeaturizer.

UnitAttributeFeaturizer::Data cherrypi::UnitAttributeFeaturizer::extract	(	State *	state,
		Rect const &	boundingBox = Rect()
	)

Extract unit features for all live units.

UnitAttributeFeaturizer::Data cherrypi::UnitAttributeFeaturizer::extract	(	State *	state,
		UnitFilter	filter,
		Rect const &	boundingBox = Rect()
	)

Extract unit features for all live units that pass the given filter.

virtual void cherrypi::UnitAttributeFeaturizer::extractUnit ( TensorDest  acc, Unit *  unit  )

protectedpure virtual

Reimplement this in actual featurizers.

This function is expected to set acc[0], ..., acc[numChannels-1]

Implemented in cherrypi::UnitStatFeaturizer, cherrypi::UnitFlagsFeaturizer, cherrypi::UnitTypeDefoggerFeaturizer, cherrypi::UnitTypeFeaturizer, and cherrypi::UnitPresenceFeaturizer.

FeatureData cherrypi::UnitAttributeFeaturizer::toSpatialFeature	(	Data const &	data,
		SubsampleMethod	pooling = SubsampleMethod::Sum
	)		const

Embeds the unit attribute data into a spatial feature.

void cherrypi::UnitAttributeFeaturizer::toSpatialFeature	(	FeatureData *	dest,
		Data const &	data,
		SubsampleMethod	pooling = SubsampleMethod::Sum
	)		const

Embeds the unit attribute data into a spatial feature.

This version will re-use the tensor memory of the given feature data instance.

Member Data Documentation

std::shared_ptr<BaseJitter> cherrypi::UnitAttributeFeaturizer::jitter = std::make_shared<NoJitter>()

Optional jittering of unit positions.

std::string cherrypi::UnitAttributeFeaturizer::name

int cherrypi::UnitAttributeFeaturizer::numChannels

CustomFeatureType cherrypi::UnitAttributeFeaturizer::type

---

The documentation for this struct was generated from the following files:

• src/features/unitsfeatures.h
• src/features/unitsfeatures.cpp