Package edu.wpi.grip.core.operations

Class CudaOperation

java.lang.Object

edu.wpi.grip.core.operations.CudaOperation

All Implemented Interfaces:

Operation

Direct Known Subclasses:

BlurOperation, CannyEdgeOperation, DesaturateOperation, NormalizeOperation

public abstract class CudaOperation
extends java.lang.Object
implements Operation

A partial implementation of Operation that has the option to use CUDA acceleration.

Field Summary

Fields

Modifier and Type	Field	Description
protected org.bytedeco.javacpp.opencv_core.Mat	cpuIn	The mat used for an input to the CPU operation.
protected org.bytedeco.javacpp.opencv_core.Mat	cpuOut	The output mat of a CPU operation.
protected SocketHint<java.lang.Boolean>	gpuHint
protected org.bytedeco.javacpp.opencv_core.GpuMat	gpuIn	The mat used for an input to the CUDA operation.
protected org.bytedeco.javacpp.opencv_core.GpuMat	gpuOut	The output mat of a CUDA operation.
protected CudaSocket	gpuSocket	Input socket telling the operation to prefer to use CUDA acceleration when possible.
protected SocketHint<MatWrapper>	inputHint
protected InputSocket<MatWrapper>	inputSocket	Default image input socket.
protected SocketHint<MatWrapper>	outputHint
protected OutputSocket<MatWrapper>	outputSocket	Default image output socket.

Constructor Summary

Constructors

Modifier	Constructor	Description
protected	CudaOperation(InputSocket.Factory isf, OutputSocket.Factory osf)

Method Summary

Modifier and Type	Method	Description
void	cleanUp()	Allows the step to clean itself up when removed from the pipeline.
protected boolean	preferCuda()	Checks the gpuSocket to see if this operation should prefer to use the CUDA codepath.

Methods inherited from class java.lang.Object

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

Methods inherited from interface edu.wpi.grip.core.Operation

getInputSockets, getOutputSockets, perform

Field Detail

inputHint

protected final SocketHint<MatWrapper> inputHint

gpuHint

protected final SocketHint<java.lang.Boolean> gpuHint

outputHint

protected final SocketHint<MatWrapper> outputHint

inputSocket

protected final InputSocket<MatWrapper> inputSocket

Default image input socket.

gpuSocket

protected final CudaSocket gpuSocket

Input socket telling the operation to prefer to use CUDA acceleration when possible.

outputSocket

protected final OutputSocket<MatWrapper> outputSocket

Default image output socket.

cpuIn

protected final org.bytedeco.javacpp.opencv_core.Mat cpuIn

The mat used for an input to the CPU operation.

gpuIn

protected final org.bytedeco.javacpp.opencv_core.GpuMat gpuIn

The mat used for an input to the CUDA operation.

cpuOut

protected final org.bytedeco.javacpp.opencv_core.Mat cpuOut

The output mat of a CPU operation.

gpuOut

protected final org.bytedeco.javacpp.opencv_core.GpuMat gpuOut

The output mat of a CUDA operation.

Constructor Detail

CudaOperation

protected CudaOperation(InputSocket.Factory isf,
                        OutputSocket.Factory osf)

Method Detail

cleanUp

public void cleanUp()

Description copied from interface: Operation

Allows the step to clean itself up when removed from the pipeline. This should only be called by Step.setRemoved() to ensure correct synchronization.

Specified by:

cleanUp in interface Operation

preferCuda

protected boolean preferCuda()

Checks the gpuSocket to see if this operation should prefer to use the CUDA codepath.

Returns:

true if this operation should prefer to use CUDA, false if it should only use the CPU