Package org.anchoranalysis.image.inference.bean.segment.instance

Class SegmentStackIntoObjectsPooled<T extends InferenceModel>

Object

org.anchoranalysis.bean.AnchorBean<B>

org.anchoranalysis.bean.initializable.InitializableBean<T,org.anchoranalysis.image.bean.nonbean.init.ImageInitialization>

org.anchoranalysis.image.bean.ImageBean<T>

org.anchoranalysis.image.bean.segment.SegmentationBean<SegmentStackIntoObjectsPooled<T>>

org.anchoranalysis.image.inference.bean.segment.instance.SegmentStackIntoObjectsPooled<T>

Type Parameters:

T - model-type

Direct Known Subclasses:

SegmentStackIntoObjectsScaleDecode, SuppressNonMaximum

public abstract class SegmentStackIntoObjectsPooled<T extends InferenceModel>
extends org.anchoranalysis.image.bean.segment.SegmentationBean<SegmentStackIntoObjectsPooled<T>>

A base class for algorithms to segment a stack into one or more objects - using a pool of models

These models are typically CNN deep-learning models.

Author:

Owen Feehan

Constructor Summary

Constructors

Constructor	Description
SegmentStackIntoObjectsPooled()

Method Summary

Modifier and Type	Method	Description
abstract ConcurrentModelPool<T>	createModelPool(ConcurrencyPlan plan, Logger logger)	Creates the model pool (to be used by multiple threads).
protected Path	resolve(String modelFilename)	Resolves a relative filename for a model into a path, relative to the model directory.
SegmentedObjects	segment(Stack stack, ExecutionTimeRecorder executionTimeRecorder)	Segments individually using a pool of size 1 just for one stack.
abstract SegmentedObjects	segment(Stack stack, ConcurrentModelPool<T> modelPool, ExecutionTimeRecorder executionTimeRecorder)	Segments a stack to produce an object-collection.

Methods inherited from class org.anchoranalysis.bean.initializable.InitializableBean

getInitialization, getLogger, getPropertyInitializer, initialize, initializeRecursive, initRecursiveWithInitializer, isInitialized, onInitialization

Methods inherited from class org.anchoranalysis.bean.AnchorBean

checkMisconfigured, describeBean, describeChildren, duplicateBean, fields, findFieldsOfClass, getBeanName, getLocalPath, localise, toString

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, wait, wait, wait

Constructor Details

SegmentStackIntoObjectsPooled

public SegmentStackIntoObjectsPooled()

Method Details

segment

public SegmentedObjects segment(Stack stack, ExecutionTimeRecorder executionTimeRecorder) throws org.anchoranalysis.image.bean.nonbean.segment.SegmentationFailedException

Segments individually using a pool of size 1 just for one stack.

See segment(Stack, ConcurrentModelPool, ExecutionTimeRecorder) for more details.

Parameters:

stack - the stack to segment.

executionTimeRecorder - for measuring execution-times of operations.

Returns:

a collection of objects with corresponding confidence scores.

Throws:

org.anchoranalysis.image.bean.nonbean.segment.SegmentationFailedException - if anything goes wrong during the segmentation.

createModelPool

public abstract ConcurrentModelPool<T> createModelPool(ConcurrencyPlan plan, Logger logger) throws CreateModelFailedException

Creates the model pool (to be used by multiple threads).

Parameters:

plan - the number and types of processors available for concurrent execution.

logger - the logger.

Returns:

the newly created model pool.

Throws:

CreateModelFailedException - if a model cannot be created.

segment

public abstract SegmentedObjects segment(Stack stack, ConcurrentModelPool<T> modelPool, ExecutionTimeRecorder executionTimeRecorder) throws org.anchoranalysis.image.bean.nonbean.segment.SegmentationFailedException

Segments a stack to produce an object-collection.

Any created objects will always exist inside the stack's Extent.

Parameters:

stack - the stack to segment.

modelPool - the pool of model instances which can each be used for inference (in parallel).

executionTimeRecorder - measures execution-times of particular operations.

Returns:

a collection of objects with corresponding confidence scores.

Throws:

org.anchoranalysis.image.bean.nonbean.segment.SegmentationFailedException - if anything goes wrong during the segmentation.

resolve

protected Path resolve(String modelFilename) throws InitializeException

Resolves a relative filename for a model into a path, relative to the model directory.

Parameters:

modelFilename - the filename for the model (to the model directory).

Returns:

an absolute path to the model.

Throws:

InitializeException - if a bean requires initialization, but has not been initialized.