.TH "MPSCNNLossLabels" 3 "Mon Jul 9 2018" "Version MetalPerformanceShaders-119.3" "MetalPerformanceShaders.framework" \" -*- nroff -*- .ad l .nh .SH NAME MPSCNNLossLabels .SH SYNOPSIS .br .PP .PP \fC#import \fP .PP Inherits \fBMPSState\fP\&. .SS "Instance Methods" .in +1c .ti -1c .RI "(nonnull instancetype) \- \fBinit\fP" .br .ti -1c .RI "(nonnull instancetype) \- \fBinitWithDevice:labelsDescriptor:\fP" .br .ti -1c .RI "(nonnull instancetype) \- \fBinitWithDevice:lossImageSize:labelsDescriptor:weightsDescriptor:\fP" .br .ti -1c .RI "(nonnull \fBMPSImage\fP *) \- \fBlossImage\fP" .br .ti -1c .RI "(nonnull \fBMPSImage\fP *) \- \fBlabelsImage\fP" .br .ti -1c .RI "(nonnull \fBMPSImage\fP *) \- \fBweightsImage\fP" .br .in -1c .SS "Additional Inherited Members" .SH "Detailed Description" .PP This depends on Metal\&.framework\&. The \fBMPSCNNLossLabels\fP is used to hold the per-element weights buffer used by both \fBMPSCNNLoss\fP forward filter and MPSCNNLossGradient backward filter\&. The \fBMPSCNNLoss\fP forward filter populates the \fBMPSCNNLossLabels\fP object and the MPSCNNLossGradient backward filter consumes the state object\&. .SH "Method Documentation" .PP .SS "\- (nonnull instancetype) init " Use one of the interfaces below instead\&. .PP Reimplemented from \fBMPSState\fP\&. .SS "\- (nonnull instancetype) initWithDevice: (nonnull id< MTLDevice >) device(\fBMPSCNNLossDataDescriptor\fP *_Nonnull) labelsDescriptor" Set labels (aka targets, ground truth) for the \fBMPSCNNLossLabels\fP object\&. The labels and weights data are copied into internal storage\&. The computed loss can either be a scalar value (in batch mode, a single value per image in a batch) or it can be one value per feature channel\&. Thus, the size of the loss image must either match the size of the input source image or be {1, 1, 1}, which results in a scalar value\&. In this convinience initializer, the assumed size of the loss image is {1, 1, 1}\&. .PP \fBParameters:\fP .RS 4 \fIdevice\fP Device the state resources will be created on\&. .br \fIlabelsDescriptor\fP Describes the labels data\&. This includes: .IP "\(bu" 2 The per-element labels data\&. The data must be in floating point format\&. .IP "\(bu" 2 Data layout of labels data\&. See MPSImage\&.h for more information\&. .IP "\(bu" 2 Size of labels data: (width, height, feature channels}\&. .IP "\(bu" 2 Optionally, row bytes of labels data\&. .IP "\(bu" 2 Optionally, slice bytes of labels data\&. .PP .RE .PP .SS "\- (nonnull instancetype) initWithDevice: (nonnull id< MTLDevice >) device(MTLSize) lossImageSize(\fBMPSCNNLossDataDescriptor\fP *_Nonnull) labelsDescriptor(\fBMPSCNNLossDataDescriptor\fP *_Nullable) weightsDescriptor" Set labels (aka targets, ground truth) and weights for the \fBMPSCNNLossLabels\fP object\&. Weights are optional\&. The labels and weights data are copied into internal storage\&. .PP \fBParameters:\fP .RS 4 \fIdevice\fP Device the state resources will be created on\&. .br \fIlossImageSize\fP The size of the resulting loss image: { width, height, featureChannels }\&. The computed loss can either be a scalar value (in batch mode, a single value per image in a batch) or it can be one value per feature channel\&. Thus, the size of the loss image must either match the size of the input source image or be {1, 1, 1}, which results in a scalar value\&. .br \fIlabelsDescriptor\fP Describes the labels data\&. This includes: .IP "\(bu" 2 The per-element labels data\&. The data must be in floating point format\&. .IP "\(bu" 2 Data layout of labels data\&. See MPSImage\&.h for more information\&. .IP "\(bu" 2 Size of labels data: (width, height, feature channels}\&. .IP "\(bu" 2 Optionally, row bytes of labels data\&. .IP "\(bu" 2 Optionally, slice bytes of labels data\&. .PP .br \fIweightsDescriptor\fP Describes the weights data\&. This includes: .IP "\(bu" 2 The per-element weights data\&. The data must be in floating point format\&. .IP "\(bu" 2 Data layout of weights data\&. See MPSImage\&.h for more information\&. .IP "\(bu" 2 Size of weights data: (width, height, feature channels}\&. .IP "\(bu" 2 Optionally, row bytes of weights data\&. .IP "\(bu" 2 Optionally, slice bytes of weights data\&. This parameter is optional\&. If you are using a single weight, please use the weight property of the \fBMPSCNNLossDescriptor\fP object\&. .PP .RE .PP .SS "\- (nonnull \fBMPSImage\fP*) labelsImage " Labels image accessor method\&. .PP \fBReturns:\fP .RS 4 An autoreleased \fBMPSImage\fP object, containing the labels data\&. The labels data is populated in the -initWithDevice call\&. .RE .PP In order to gaurantee that the image is correctly synchronized for CPU side access, it is the application's responsibility to call the [gradientState synchronizeOnCommandBuffer:] method before accessing the data in the image\&. .SS "\- (nonnull \fBMPSImage\fP*) lossImage " Loss image accessor method\&. .PP \fBReturns:\fP .RS 4 An autoreleased \fBMPSImage\fP object, containing the loss data\&. The loss data is populated in the -encode call, thus the contents are undefined until you -encode the filter\&. .RE .PP In order to gaurantee that the image is correctly synchronized for CPU side access, it is the application's responsibility to call the [gradientState synchronizeOnCommandBuffer:] method before accessing the data in the image\&. .SS "\- (nonnull \fBMPSImage\fP*) weightsImage " Weights image accessor method\&. .PP \fBReturns:\fP .RS 4 An autoreleased \fBMPSImage\fP object, containing the weights data\&. The weights data is populated in the -initWithDevice call\&. .RE .PP In order to gaurantee that the image is correctly synchronized for CPU side access, it is the application's responsibility to call the [gradientState synchronizeOnCommandBuffer:] method before accessing the data in the image\&. .SH "Author" .PP Generated automatically by Doxygen for MetalPerformanceShaders\&.framework from the source code\&.