Class InferenceModelConfiguration
Definition
- Namespace:
- Tizen.Multimedia.Vision
- Assembly:
- Tizen.Multimedia.Vision.dll
Represents a configuration of FaceDetector, FacialLandmarkDetector, ImageClassifier and ObjectDetector.
public class InferenceModelConfiguration : EngineConfiguration
- Inheritance
Remarks
'Inference model' means pre-learned data, which is represented by ConfigurationFilePath and
WeightFilePath, CategoryFilePath.
If user want to use tizen default inference model and its related value,
Please refer Tizen guide page(https://developer.tizen.org/development/guides/.net-application).
Constructors
View SourceInferenceModelConfiguration()
Initializes a new instance of the InferenceModelConfiguration class.
Declaration
public InferenceModelConfiguration()
Remarks
'Inference model' means pre-learned data, which is represented by ConfigurationFilePath and
WeightFilePath, CategoryFilePath.
If user want to use tizen default inference model and its related value,
Please refer Tizen guide page(https://developer.tizen.org/development/guides/.net-application).
Properties
View SourceBackend
Gets or sets the inference model's backend engine.
Declaration
public InferenceBackendType Backend { get; set; }
Property Value
Type | Description |
---|---|
InferenceBackendType |
Remarks
The default backend type is OpenCV
See Also
View SourceCategoryFilePath
Gets or sets the path of inference model's category file.
Declaration
public string CategoryFilePath { get; set; }
Property Value
Type | Description |
---|---|
string |
Remarks
This value should be set to use ImageClassifier or ObjectDetector.
ConfidenceThreshold
Gets or sets the threshold of confidence.
Declaration
public double ConfidenceThreshold { get; set; }
Property Value
Type | Description |
---|---|
double |
Remarks
The vaild range is greater than or equal to 0.0 and less than or equal to 1.0.
The value 1.0 means maximum accuracy.
ConfigurationFilePath
Gets or sets the path of inference model's configuration data file.
Declaration
public string ConfigurationFilePath { get; set; }
Property Value
Type | Description |
---|---|
string |
Remarks
'Inference model' means pre-learned data, which is represented by ConfigurationFilePath and
WeightFilePath, CategoryFilePath.
If user want to use tizen default inference model and its related value,
Please refer Tizen guide page(https://developer.tizen.org/development/guides/.net-application).
DataType
Gets or sets the type of data used for inference model.
Declaration
public InferenceDataType DataType { get; set; }
Property Value
Type | Description |
---|---|
InferenceDataType |
Remarks
For example, this value should be set to Float32 for a model data supporting float32.
Float32 will be used internally if a user doesn't set the value.
Device
Gets or sets the processor type for inference models.
Declaration
public InferenceTargetDevice Device { get; set; }
Property Value
Type | Description |
---|---|
InferenceTargetDevice |
Remarks
The default device is CPU.
If a device doesn't support GPU and Custom,
CPU will be used internally, despite the user's choice.
InputNodeName
Gets or sets the name of an input node
Declaration
public string InputNodeName { get; set; }
Property Value
Type | Description |
---|---|
string |
Remarks
'Inference model' means pre-learned data, which is represented by ConfigurationFilePath and
WeightFilePath, CategoryFilePath.
If user want to use tizen default inference model and its related value,
Please refer Tizen guide page(https://developer.tizen.org/development/guides/.net-application).
MaxOutputNumber
Gets or sets the maximum output number of detection or classification.
Declaration
public int MaxOutputNumber { get; set; }
Property Value
Type | Description |
---|---|
int |
Remarks
The input value over 10 will be set to 10 and the input value under 1 will be set to 1.
This value can be used to decide the size of Roi, it's length should be the same.
MeanValue
Gets or sets the inference model's mean value.
Declaration
public double MeanValue { get; set; }
Property Value
Type | Description |
---|---|
double |
Remarks
It should be greater than or equal to 0.
MetadataFilePath
Gets or sets the path of inference model's metadata file.
Declaration
public string MetadataFilePath { get; set; }
Property Value
Type | Description |
---|---|
string |
Remarks
This value should be set to use ImageClassifier or ObjectDetector.
OutputNodeName
Gets or sets the name of an output node
Declaration
public IList<string> OutputNodeName { get; set; }
Property Value
Type | Description |
---|---|
IList<><string> |
Remarks
'Inference model' means pre-learned data, which is represented by ConfigurationFilePath and
WeightFilePath, CategoryFilePath.
If user want to use tizen default inference model and its related value,
Please refer Tizen guide page(https://developer.tizen.org/development/guides/.net-application).
Roi
Gets or sets the ROI(Region Of Interest) of ImageClassifier and FacialLandmarkDetector
Declaration
public Rectangle? Roi { get; set; }
Property Value
Type | Description |
---|---|
Rectangle? |
Remarks
Default value is null. If Roi is null, the entire region of MediaVisionSource will be analyzed.
See Also
View SourceStdValue
Gets or sets the inference model's STD(Standard deviation) value.
Declaration
public double StdValue { get; set; }
Property Value
Type | Description |
---|---|
double |
Remarks
It should be greater than or equal to 0.
SupportedBackend
Gets the list of inference backend engine which is supported in the current device.
Declaration
public IEnumerable<InferenceBackendType> SupportedBackend { get; }
Property Value
Type | Description |
---|---|
IEnumerable<><InferenceBackendType> | If there's no supported backend, empty collection will be returned. |
Remarks
'Inference model' means pre-learned data, which is represented by ConfigurationFilePath and
WeightFilePath, CategoryFilePath.
If user want to use tizen default inference model and its related value,
Please refer Tizen guide page(https://developer.tizen.org/development/guides/.net-application).
TensorChannels
Gets or sets the number of inference model's tensor channel.
Declaration
public int TensorChannels { get; set; }
Property Value
Type | Description |
---|---|
int |
Remarks
For example, for RGB colorspace this value should be set to 3
It should be greater than 0.
TensorSize
Gets or sets the size of inference model's tensor.
Declaration
public Size TensorSize { get; set; }
Property Value
Type | Description |
---|---|
Size |
Remarks
Both width and height of tensor should be greater than 0.
'Size(-1, -1) is allowed when the intention is to use original image source size as TensorSize.
WeightFilePath
Gets or sets the path of inference model's weight file.
Declaration
public string WeightFilePath { get; set; }
Property Value
Type | Description |
---|---|
string |
Remarks
'Inference model' means pre-learned data, which is represented by ConfigurationFilePath and
WeightFilePath, CategoryFilePath.
If user want to use tizen default inference model and its related value,
Please refer Tizen guide page(https://developer.tizen.org/development/guides/.net-application).
Methods
View SourceDispose(bool)
Releases the resources used by the InferenceModelConfiguration object.
Declaration
protected override void Dispose(bool disposing)
Parameters
Type | Name | Description |
---|---|---|
bool | disposing | true to release both managed and unmanaged resources, otherwise false to release only unmanaged resources. |
Overrides
Remarks
'Inference model' means pre-learned data, which is represented by ConfigurationFilePath and
WeightFilePath, CategoryFilePath.
If user want to use tizen default inference model and its related value,
Please refer Tizen guide page(https://developer.tizen.org/development/guides/.net-application).
LoadInferenceModel()
Loads inference model data and its related attributes.
Declaration
public void LoadInferenceModel()
Remarks
Before calling this method, user should set all properties which is required by each inference model.
The properties set after calling this method will not be affected in the result.
Exceptions
Type | Condition |
---|---|
FileFormatException | Invalid data type is used in inference model data. |