OpenCV5Sharp

Appearance

DNN Module API Reference ​

Complete documentation for the DNN classes, methods, and enums in OpenCV5Sharp. Equivalent to the Official OpenCV Dnn Documentation.

📦 Classes and Structs ​

DnnClassificationModel ​

Inherits from: DnnModel

This class represents high-level API for classification models. * * ClassificationModel allows to set params for preprocessing input image. * ClassificationModel creates net from file with trained weights and config, * sets preprocessing input, runs forward pass and return top-1 prediction.

Constructors ​

  • new DnnClassificationModel(string model, string? config)
    • Summary: * Create classification model from network represented in one of the supported formats.
    • Parameter model: The model parameter.
    • Parameter config: The config parameter.
  • new DnnClassificationModel(DnnNet network)
    • Summary: * Create model from deep learning network.
    • Parameter network: The network parameter.

Methods ​

  • DnnClassificationModel? SetEnableSoftmaxPostProcessing(bool enable)

    • Summary: * Set enable/disable softmax post processing option.
    • Remarks:

  • If this option is true, softmax is applied after forward inference within the classify() function

  • to convert the confidences range to [0.0-1.0].

  • This function allows you to toggle this behavior.

  • Please turn true when not contain softmax layer in model.

    • Parameter enable: Set enable softmax post processing within the classify() function.

    • Parameter enable: The enable parameter.

    • Returns: The returned value.

  • bool GetEnableSoftmaxPostProcessing()

    • Summary: * Get enable/disable softmax post processing option.
    • Remarks:

  • This option defaults to false, softmax post processing is not applied within the classify() function.

    • Returns: The returned value.

  • void Classify(Mat frame, int classId, float conf)

    • Summary: This is an overloaded member function, provided for convenience.
    • Parameter frame: The frame parameter.
    • Parameter classId: The classId parameter.
    • Parameter conf: The conf parameter.

DnnDetectionModel ​

Inherits from: DnnModel

This class represents high-level API for object detection networks. * * DetectionModel allows to set params for preprocessing input image. * DetectionModel creates net from file with trained weights and config, * sets preprocessing input, runs forward pass and return result detections. * For DetectionModel SSD, Faster R-CNN, YOLO topologies are supported.

Constructors ​

  • new DnnDetectionModel(string model, string? config)
    • Summary: * Create detection model from network represented in one of the supported formats.
    • Parameter model: The model parameter.
    • Parameter config: The config parameter.
  • new DnnDetectionModel(DnnNet network)
    • Summary: * Create model from deep learning network.
    • Parameter network: The network parameter.

Methods ​

  • DnnDetectionModel? SetNmsAcrossClasses(bool value)

    • Summary: * nmsAcrossClasses defaults to false,
    • Remarks:

  • such that when non max suppression is used during the detect() function, it will do so per-class.

  • This function allows you to toggle this behaviour.

    • Parameter value: The new value for nmsAcrossClasses

    • Parameter value: The value parameter.

    • Returns: The returned value.

  • bool GetNmsAcrossClasses()

    • Summary: * Getter for nmsAcrossClasses. This variable defaults to false,
    • Remarks:

  • such that when non max suppression is used during the detect() function, it will do so only per-class

    • Returns: The returned value.

  • void Detect(Mat frame, IntPtr classIds, IntPtr confidences, IntPtr boxes, float confThreshold, float nmsThreshold)

    • Summary: Given the input frame, create input blob, run net and return result detections. * frame The input image. * classIds Class indexes in result detection. * confidences A set of corresponding confidences. * boxes A set of bounding boxes. * confThreshold A threshold used to filter boxes by confidences. * nmsThreshold A threshold used in non maximum suppression.
    • Parameter frame: The frame parameter.
    • Parameter classIds: The classIds parameter.
    • Parameter confidences: The confidences parameter.
    • Parameter boxes: The boxes parameter.
    • Parameter confThreshold: The confThreshold parameter.
    • Parameter nmsThreshold: The nmsThreshold parameter.

DnnDict ​

Inherits from: DisposableOpenCVObject

Stub class representing Dnn.Dict.

Constructors ​

  • new DnnDict()
    • Summary: Creates a new empty DnnDict instance.

DnnDictValue ​

Inherits from: DisposableOpenCVObject

This struct stores the scalar value (or array) of one of the following type: double, String or int64. *

Constructors ​

  • new DnnDictValue(int i)
    • Summary: Creates a DnnDictValue from an integer value.
    • Parameter i: The integer value to store.
  • new DnnDictValue(double p)
    • Summary: Creates a DnnDictValue from a double (real) value.
    • Parameter p: The double value to store.
  • new DnnDictValue(string s)
    • Summary: Creates a DnnDictValue from a string value.
    • Parameter s: The string value to store.

Methods ​

  • bool IsInt()
    • Summary: Checks whether the stored value is of integer type.
    • Returns: true if the stored value is an integer; otherwise false.
  • bool IsString()
    • Summary: Checks whether the stored value is of string type.
    • Returns: true if the stored value is a string; otherwise false.
  • bool IsReal()
    • Summary: Checks whether the stored value is of real (double) type.
    • Returns: true if the stored value is a real number; otherwise false.
  • int GetIntValue(int idx)
    • Summary: Returns the stored value as an integer.
    • Parameter idx: Index of the value in the array (use 0 for scalar values).
    • Returns: The integer representation of the stored value at the given index.
  • double GetRealValue(int idx)
    • Summary: Returns the stored value as a double.
    • Parameter idx: Index of the value in the array (use 0 for scalar values).
    • Returns: The double representation of the stored value at the given index.
  • string? GetStringValue(int idx)
    • Summary: Returns the stored value as a string.
    • Parameter idx: Index of the value in the array (use 0 for scalar values).
    • Returns: The string representation of the stored value at the given index, or null.

DnnImage2BlobParams ​

Inherits from: DisposableOpenCVObject

Processing params of image to blob. * * It includes all possible image processing operations and corresponding parameters. * * See also: blobFromImageWithParams * * Note: * The order and usage of scalefactor and mean are (input - mean) * scalefactor. * The order and usage of scalefactor, size, mean, swapRB, and ddepth are consistent * with the function of blobFromImage.

Properties ​

PropertyTypeDescription
ScalefactorScalarGets or sets the scalefactor property.
SizeSizeGets or sets the size property.
MeanScalarGets or sets the mean property.
SwapRBboolGets or sets the swapRB property.
DdepthintGets or sets the ddepth property.
DatalayoutDataLayoutGets or sets the datalayout property.
PaddingmodeIntPtrGets or sets the paddingmode property.
BorderValueScalarGets or sets the borderValue property.

Constructors ​

  • new DnnImage2BlobParams()
    • Summary: Creates a DnnImage2BlobParams with default preprocessing parameters.
  • new DnnImage2BlobParams(Scalar scalefactor, Size size, Scalar mean, bool swapRB, int ddepth, DataLayout datalayout, DnnImagePaddingMode mode, Scalar borderValue)
    • Summary: Creates a DnnImage2BlobParams with the specified preprocessing parameters.
    • Parameter scalefactor: The scalefactor parameter.
    • Parameter size: The size parameter.
    • Parameter mean: The mean parameter.
    • Parameter swapRB: The swapRB parameter.
    • Parameter ddepth: The ddepth parameter.
    • Parameter datalayout: The datalayout parameter.
    • Parameter mode: The mode parameter.
    • Parameter borderValue: The borderValue parameter.

Methods ​

  • Rect BlobRectToImageRect(Rect rBlob, Size size)
    • Summary: Get rectangle coordinates in original image system from rectangle in blob coordinates. * rBlob rect in blob coordinates. * size original input image size. * Returns: rectangle in original image coordinates.
    • Parameter rBlob: The rBlob parameter.
    • Parameter size: The size parameter.
    • Returns: The returned value.
  • void BlobRectsToImageRects(IntPtr rBlob, IntPtr rImg, Size size)
    • Summary: Get rectangle coordinates in original image system from rectangle in blob coordinates. * rBlob rect in blob coordinates. * rImg result rect in image coordinates. * size original input image size.
    • Parameter rBlob: The rBlob parameter.
    • Parameter rImg: The rImg parameter.
    • Parameter size: The size parameter.

DnnKeypointsModel ​

Inherits from: DnnModel

This class represents high-level API for keypoints models * * KeypointsModel allows to set params for preprocessing input image. * KeypointsModel creates net from file with trained weights and config, * sets preprocessing input, runs forward pass and returns the x and y coordinates of each detected keypoint

Constructors ​

  • new DnnKeypointsModel(string model, string? config)
    • Summary: * Create keypoints model from network represented in one of the supported formats.
    • Parameter model: The model parameter.
    • Parameter config: The config parameter.
  • new DnnKeypointsModel(DnnNet network)
    • Summary: * Create model from deep learning network.
    • Parameter network: The network parameter.

Methods ​

  • IntPtr Estimate(Mat frame, float thresh)
    • Summary: Given the input frame, create input blob, run net * frame The input image. * thresh minimum confidence threshold to select a keypoint * Returns: a vector holding the x and y coordinates of each detected keypoint *
    • Parameter frame: The frame parameter.
    • Parameter thresh: The thresh parameter.
    • Returns: The returned value.

DnnLayer ​

Inherits from: Algorithm

This interface class allows to build new Layers - are building blocks of networks. * * Each class, derived from Layer, must implement forward() method to compute outputs. * Also before using the new layer into networks you must register your layer by using one of LayerFactory macros.

Properties ​

PropertyTypeDescription
BlobsMat[]Gets or sets the blobs property.
Namestring?Gets or sets the name property.
Typestring?Gets or sets the type property.
PreferableTargetintGets or sets the preferableTarget property.

Methods ​

  • void Finalize(IntPtr inputs, IntPtr outputs)
    • Summary: Computes and sets internal parameters according to inputs, outputs and blobs. * inputs vector of already allocated input blobs * outputs vector of already allocated output blobs * * This method is called after network has allocated all memory for input and output blobs * and before inferencing.
    • Parameter inputs: The inputs parameter.
    • Parameter outputs: The outputs parameter.
  • void Run(IntPtr inputs, IntPtr outputs, IntPtr internals)
    • Summary: Allocates layer and computes output. * (Deprecated) This method will be removed in the future release.
    • Parameter inputs: The inputs parameter.
    • Parameter outputs: The outputs parameter.
    • Parameter internals: The internals parameter.
  • int OutputNameToIndex(string outputName)
    • Summary: Returns index of output blob in output array. * See also: inputNameToIndex()
    • Parameter outputName: The outputName parameter.
    • Returns: The returned value.

DnnLayerParams ​

Inherits from: DisposableOpenCVObject

Stub class representing Dnn.LayerParams.

Constructors ​

  • new DnnLayerParams()
    • Summary: Creates a new empty DnnLayerParams instance for configuring layer parameters.

DnnModel ​

Inherits from: DisposableOpenCVObject

This class is presented high-level API for neural networks. * * Model allows to set params for preprocessing input image. * Model creates net from file with trained weights and config, * sets preprocessing input and runs forward pass.

Constructors ​

  • new DnnModel(string model, string? config)
    • Summary: * Create model from deep learning network represented in one of the supported formats.
    • Parameter model: The model parameter.
    • Parameter config: The config parameter.
  • new DnnModel(DnnNet network)
    • Summary: * Create model from deep learning network.
    • Parameter network: The network parameter.

Methods ​

  • DnnModel? SetInputSize(Size size)

    • Summary: Set input size for frame. * size New input size. * Note: If shape of the new blob less than 0, then frame size not change.
    • Parameter size: The size parameter.
    • Returns: The returned value.

  • DnnModel? SetInputSize(int width, int height)

    • Summary: This is an overloaded member function, provided for convenience.
    • Remarks:
    • Parameter width: New input width.
    • Parameter height: New input height.

  • Parameter width: The width parameter.

  • Parameter height: The height parameter.

  • Returns: The returned value.

  • DnnModel? SetInputMean(Scalar mean)

    • Summary: Set mean value for frame. * mean Scalar with mean values which are subtracted from channels.
    • Parameter mean: The mean parameter.
    • Returns: The returned value.

  • DnnModel? SetInputScale(Scalar scale)

    • Summary: Set scalefactor value for frame. * scale Multiplier for frame values.
    • Parameter scale: The scale parameter.
    • Returns: The returned value.

  • DnnModel? SetInputCrop(bool crop)

    • Summary: Set flag crop for frame. * crop Flag which indicates whether image will be cropped after resize or not.
    • Parameter crop: The crop parameter.
    • Returns: The returned value.

  • DnnModel? SetInputSwapRB(bool swapRB)

    • Summary: Set flag swapRB for frame. * swapRB Flag which indicates that swap first and last channels.
    • Parameter swapRB: The swapRB parameter.
    • Returns: The returned value.

  • DnnModel? SetOutputNames(IntPtr outNames)

    • Summary: Set output names for frame. * outNames Names for output layers.
    • Parameter outNames: The outNames parameter.
    • Returns: The returned value.

  • void SetInputParams(double scale, Size size, Scalar mean, bool swapRB, bool crop)

    • Summary: Set preprocessing parameters for frame. * size New input size. * mean Scalar with mean values which are subtracted from channels. * scale Multiplier for frame values. * swapRB Flag which indicates that swap first and last channels. * crop Flag which indicates whether image will be cropped after resize or not. * blob(n, c, y, x) = scale * resize( frame(y, x, c) ) - mean(c) )
    • Parameter scale: The scale parameter.
    • Parameter size: The size parameter.
    • Parameter mean: The mean parameter.
    • Parameter swapRB: The swapRB parameter.
    • Parameter crop: The crop parameter.

  • void Predict(Mat frame, IntPtr outs)

    • Summary: Given the input frame, create input blob, run net and return the output blobs. * frame The input image. * outs Allocated output blobs, which will store results of the computation.
    • Parameter frame: The frame parameter.
    • Parameter outs: The outs parameter.

  • DnnModel? SetPreferableBackend(IntPtr backendId)

    • Summary: Sets the preferred computation backend for the model.
    • Parameter backendId: Backend identifier (see DnnBackend enum).
    • Returns: The DnnModel instance for method chaining, or null.

  • DnnModel? SetPreferableTarget(IntPtr targetId)

    • Summary: Sets the preferred target device for the model.
    • Parameter targetId: Target identifier (see DnnTarget enum).
    • Returns: The DnnModel instance for method chaining, or null.

  • DnnModel? EnableWinograd(bool useWinograd)

    • Summary: Enables or disables the Winograd compute branch for 3x3 convolutions. Can speed up inference at a small loss of accuracy.
    • Parameter useWinograd: true to enable the Winograd compute branch; false to disable. Default is true.
    • Returns: The DnnModel instance for method chaining, or null.

DnnNet ​

Inherits from: DisposableOpenCVObject

This class allows to create and manipulate comprehensive artificial neural networks. * * Neural network is presented as directed acyclic graph (DAG), where vertices are Layer instances, * and edges specify relationships between layers inputs and outputs. * * Each network layer has unique integer id and unique string name inside its network. * LayerId can store either layer name or layer id. * * This class supports reference counting of its instances, i. e. copies point to the same instance.

Constructors ​

  • new DnnNet()
    • Summary: Creates a new empty DnnNet instance with no layers.

Methods ​

  • DnnNet? ReadFromModelOptimizer(string xml, string bin)

    • Summary: Create a network from Intel's Model Optimizer intermediate representation (IR). * xml XML configuration file with network's topology. * bin Binary file with trained weights. * Networks imported from Intel's Model Optimizer are launched in Intel's Inference Engine * backend.
    • Parameter xml: The xml parameter.
    • Parameter bin: The bin parameter.
    • Returns: The returned value.

  • DnnNet? ReadFromModelOptimizer(IntPtr bufferModelConfig, IntPtr bufferWeights)

    • Summary: Create a network from Intel's Model Optimizer in-memory buffers with intermediate representation (IR). * bufferModelConfig buffer with model's configuration. * bufferWeights buffer with model's trained weights. * Returns: Net object.
    • Parameter bufferModelConfig: The bufferModelConfig parameter.
    • Parameter bufferWeights: The bufferWeights parameter.
    • Returns: The returned value.

  • bool Empty()

    • Summary: Returns true if there are no layers in the network.
    • Returns: The returned value.

  • string? Dump()

    • Summary: Dump net to String * Returns: String with structure, hyperparameters, backend, target and fusion * Call method after setInput(). To see correct backend, target and fusion run after forward().
    • Returns: The returned value.

  • void DumpToFile(string path)

    • Summary: Dump net structure, hyperparameters, backend, target and fusion to dot file * path path to output file with .dot extension * See also: dump()
    • Parameter path: The path parameter.

  • void DumpToPbtxt(string path)

    • Summary: Dump net structure, hyperparameters, backend, target and fusion to pbtxt file * path path to output file with .pbtxt extension * * Use Netron (https://netron.app) to open the target file to visualize the model. * Call method after setInput(). To see correct backend, target and fusion run after forward().
    • Parameter path: The path parameter.

  • int AddLayer(string name, string type, int dtype, DnnLayerParams @params)

    • Summary: Adds new layer to the net. * name unique name of the adding layer. * type typename of the adding layer (type must be registered in LayerRegister). * dtype datatype of output blobs. * params parameters which will be used to initialize the creating layer. * Returns: unique identifier of created layer, or -1 if a failure will happen.
    • Parameter name: The name parameter.
    • Parameter type: The type parameter.
    • Parameter dtype: The dtype parameter.
    • Parameter params: The @params parameter.
    • Returns: The returned value.

  • int AddLayerToPrev(string name, string type, int dtype, DnnLayerParams @params)

    • Summary: Adds new layer and connects its first input to the first output of previously added layer. * See also: addLayer()
    • Parameter name: The name parameter.
    • Parameter type: The type parameter.
    • Parameter dtype: The dtype parameter.
    • Parameter params: The @params parameter.
    • Returns: The returned value.

  • int GetLayerId(string layer)

    • Summary: Converts string name of the layer to the integer identifier. * Returns: id of the layer, or -1 if the layer wasn't found.
    • Parameter layer: The layer parameter.
    • Returns: The returned value.

  • IntPtr GetLayerNames()

    • Summary: Returns the names of all layers in the network.
    • Returns: A pointer to a vector of layer name strings.

  • DnnLayer? GetLayer(int layerId)

    • Summary: Returns pointer to layer with specified id or name which the network use.
    • Parameter layerId: The layerId parameter.
    • Returns: The returned value.

  • DnnLayer? GetLayer(string layerName)

    • Summary: This is an overloaded member function, provided for convenience.
    • Remarks:

  • (Deprecated) Use int GetLayerId(string layer)

  • Parameter layerName: The layerName parameter.

  • Returns: The returned value.

  • DnnLayer? GetLayer(IntPtr layerId)

    • Summary: This is an overloaded member function, provided for convenience.
    • Remarks:

  • (Deprecated) to be removed

  • Parameter layerId: The layerId parameter.

  • Returns: The returned value.

  • void Connect(string outPin, string inpPin)

    • Summary: Connects output of the first layer to input of the second layer. * outPin descriptor of the first layer output. * inpPin descriptor of the second layer input. * * Descriptors have the following template <layer_name>[.input_number]: * - the first part of the template layer_name is string name of the added layer. * If this part is empty then the network input pseudo layer will be used; * - the second optional part of the template input_number * is either number of the layer input, either label one. * If this part is omitted then the first layer input will be used. * * See also: setNetInputs(), Layer.inputNameToIndex(), Layer.outputNameToIndex()
    • Parameter outPin: The outPin parameter.
    • Parameter inpPin: The inpPin parameter.

  • int RegisterOutput(string outputName, int layerId, int outputPort)

    • Summary: Registers network output with name * * Function may create additional 'Identity' layer. * * outputName identifier of the output * layerId identifier of the second layer * outputPort number of the second layer input * * Returns: index of bound layer (the same as layerId or newly created)
    • Parameter outputName: The outputName parameter.
    • Parameter layerId: The layerId parameter.
    • Parameter outputPort: The outputPort parameter.
    • Returns: The returned value.

  • void SetInputsNames(IntPtr inputBlobNames)

    • Summary: Sets outputs names of the network input pseudo layer. * * Each net always has special own the network input pseudo layer with id=0. * This layer stores the user blobs only and don't make any computations. * In fact, this layer provides the only way to pass user data into the network. * As any other layer, this layer can label its outputs and this function provides an easy way to do this.
    • Parameter inputBlobNames: The inputBlobNames parameter.

  • void SetInputShape(string inputName, MatShape shape)

    • Summary: Specify shape of network input.
    • Parameter inputName: The inputName parameter.
    • Parameter shape: The shape parameter.

  • Mat? Forward(string? outputName)

    • Summary: Runs forward pass to compute output of layer with name outputName. * outputName name for layer which output is needed to get Returns: blob for first output of specified layer. * By default runs forward pass for the whole network.
    • Parameter outputName: The outputName parameter.
    • Returns: The returned value.

  • AsyncArray? ForwardAsync(string? outputName)

    • Summary: Runs forward pass to compute output of layer with name outputName. * outputName name for layer which output is needed to get * By default runs forward pass for the whole network. * * This is an asynchronous version of Forward(string). * Dnn.Backend.InferenceEngine backend is required.
    • Parameter outputName: The outputName parameter.
    • Returns: The returned value.

  • void Forward(IntPtr outputBlobs, string? outputName)

    • Summary: Runs forward pass to compute output of layer with name outputName. * outputBlobs contains all output blobs for specified layer. * outputName name for layer which output is needed to get * If outputName is empty, runs forward pass for the whole network.
    • Parameter outputBlobs: The outputBlobs parameter.
    • Parameter outputName: The outputName parameter.

  • void Forward(IntPtr outputBlobs, IntPtr outBlobNames)

    • Summary: Runs forward pass to compute outputs of layers listed in outBlobNames. * outputBlobs contains blobs for first outputs of specified layers. * outBlobNames names for layers which outputs are needed to get
    • Parameter outputBlobs: The outputBlobs parameter.
    • Parameter outBlobNames: The outBlobNames parameter.

  • void SetPreferableBackend(int backendId)

    • Summary: * Ask network to use specific computation backend where it supported.
    • Remarks:

    • Parameter backendId: backend identifier. See also: * Backend

    • Parameter backendId: The backendId parameter.

  • void SetPreferableTarget(int targetId)

    • Summary: * Ask network to make computations on specific target device.
    • Remarks:
    • Parameter targetId: target identifier. See also: * Target

  • List of supported combinations backend / target:

  • | | DNN_BACKEND_OPENCV | DNN_BACKEND_INFERENCE_ENGINE | DNN_BACKEND_CUDA |

  • |------------------------|--------------------|------------------------------|-------------------|

  • | DNN_TARGET_CPU | + | + | |

  • | DNN_TARGET_OPENCL | + | + | |

  • | DNN_TARGET_OPENCL_FP16 | + | + | |

  • | DNN_TARGET_MYRIAD | | + | |

  • | DNN_TARGET_FPGA | | + | |

  • | DNN_TARGET_CUDA | | | + |

  • | DNN_TARGET_CUDA_FP16 | | | + |

  • | DNN_TARGET_HDDL | | + | |

    • Parameter targetId: The targetId parameter.

  • void FinalizeNet()

    • Summary: Finalizes the network configuration and prepares it for inference. * * This method must be called after setting backend/target via * setPreferableBackend() and setPreferableTarget(), and before the first * forward() call. It creates the underlying execution session (e.g. ONNX * Runtime session) on the configured backend/target. If not called * explicitly, the first forward() will call it automatically. * * Calling finalizeNet() early lets you pay the one-time setup cost at a * predictable point and catch configuration errors before inference.

  • void SetTracingMode(DnnTracingMode tracingMode)

    • Summary: * Set the tracing mode
    • Remarks:

    • Parameter tracingMode: the tracing mode, see DNN_TRACE_*

    • Parameter tracingMode: The tracingMode parameter.

  • DnnTracingMode GetTracingMode()

    • Summary: * Retrieve the current tracing mode
    • Returns: The returned value.

  • void SetProfilingMode(DnnProfilingMode profilingMode)

    • Summary: * Set the profiling mode
    • Remarks:

    • Parameter profilingMode: the profiling mode, see DNN_PROFILE_*

    • Parameter profilingMode: The profilingMode parameter.

  • DnnProfilingMode GetProfilingMode()

    • Summary: * Retrieve the current profiling mode
    • Returns: The returned value.

  • DnnModelFormat GetModelFormat()

    • Summary: * Retrieve the current model format, see DNN_MODEL_*
    • Returns: The returned value.

  • void SetInput(Mat blob, string? name, double scalefactor, Scalar mean)

    • Summary: Sets the new input value for the network * blob A new blob. Should have CV_32F or CV_8U depth. * name A name of input layer. * scalefactor An optional normalization scale. * mean An optional mean subtraction values. * See also: connect(String, String) to know format of the descriptor. * * If scale or mean values are specified, a final input blob is computed * as: * [see mathematical formula in OpenCV docs]
    • Parameter blob: The blob parameter.
    • Parameter name: The name parameter.
    • Parameter scalefactor: The scalefactor parameter.
    • Parameter mean: The mean parameter.

  • void SetParam(int layer, int numParam, Mat blob)

    • Summary: Sets the new value for the learned param of the layer. * layer name or id of the layer. * numParam index of the layer parameter in the Layer.blobs array. * blob the new value. * See also: Layer.blobs * Note: If shape of the new blob differs from the previous shape, * then the following forward pass may fail.
    • Parameter layer: The layer parameter.
    • Parameter numParam: The numParam parameter.
    • Parameter blob: The blob parameter.

  • void SetParam(string layerName, int numParam, Mat blob)

    • Summary: Sets the parameter blob of a layer identified by its name or output tensor name. * layerName layer name (classic engine) or raw ONNX output tensor name (ENGINE_NEW). * numParam index of the constant weight input to update (0 = kernel, 1 = bias, etc.). * blob the new parameter value.
    • Parameter layerName: The layerName parameter.
    • Parameter numParam: The numParam parameter.
    • Parameter blob: The blob parameter.

  • Mat? GetParam(int layer, int numParam)

    • Summary: Returns parameter blob of the layer. * layer name or id of the layer. * numParam index of the layer parameter in the Layer.blobs array. * See also: Layer.blobs
    • Parameter layer: The layer parameter.
    • Parameter numParam: The numParam parameter.
    • Returns: The returned value.

  • Mat? GetParam(string layerName, int numParam)

    • Summary: Returns the parameter blob of a layer identified by its name. Equivalent to GetParam(int, int) but uses a string layer name.
    • Parameter layerName: The name of the layer.
    • Parameter numParam: Index of the layer parameter in the Layer.blobs array.
    • Returns: The parameter Mat blob, or null if not found.

  • IntPtr GetUnconnectedOutLayers()

    • Summary: Returns indexes of layers with unconnected outputs. * * FIXIT: Rework API to registerOutput() approach, deprecate this call
    • Returns: The returned value.

  • IntPtr GetUnconnectedOutLayersNames()

    • Summary: Returns names of layers with unconnected outputs. * * FIXIT: Rework API to registerOutput() approach, deprecate this call
    • Returns: The returned value.

  • void GetLayerShapes(IntPtr netInputShapes, IntPtr netInputTypes, int layerId, IntPtr inLayerShapes, IntPtr outLayerShapes)

    • Summary: This is an overloaded member function, provided for convenience.
    • Remarks:

  • The only overload of getLayerShapes that should be kept in 5.x

    • Parameter netInputShapes: The netInputShapes parameter.
    • Parameter netInputTypes: The netInputTypes parameter.
    • Parameter layerId: The layerId parameter.
    • Parameter inLayerShapes: The inLayerShapes parameter.
    • Parameter outLayerShapes: The outLayerShapes parameter.

  • long GetFLOPS(IntPtr netInputShapes, IntPtr netInputTypes)

    • Summary: Computes FLOP for whole loaded model with specified input shapes. * netInputShapes vector of shapes for all net inputs. * netInputTypes vector of types for all net inputs. * Returns: computed FLOP.
    • Parameter netInputShapes: The netInputShapes parameter.
    • Parameter netInputTypes: The netInputTypes parameter.
    • Returns: The returned value.

  • void GetLayerTypes(IntPtr layersTypes)

    • Summary: Returns list of types for layer used in model. * layersTypes output parameter for returning types.
    • Parameter layersTypes: The layersTypes parameter.

  • int GetLayersCount(string layerType)

    • Summary: Returns count of layers of specified type. * layerType type. * Returns: count of layers
    • Parameter layerType: The layerType parameter.
    • Returns: The returned value.

  • void GetMemoryConsumption(IntPtr netInputShapes, IntPtr netInputTypes, long weights, long blobs)

    • Summary: Computes bytes number which are required to store * all weights and intermediate blobs for model. * netInputShapes vector of shapes for all net inputs. * netInputTypes vector of types for all net inputs. * weights output parameter to store resulting bytes for weights. * blobs output parameter to store resulting bytes for intermediate blobs.
    • Parameter netInputShapes: The netInputShapes parameter.
    • Parameter netInputTypes: The netInputTypes parameter.
    • Parameter weights: The weights parameter.
    • Parameter blobs: The blobs parameter.

  • void EnableFusion(bool fusion)

    • Summary: Enables or disables layer fusion in the network. * fusion true to enable the fusion, false to disable. The fusion is enabled by default.
    • Parameter fusion: The fusion parameter.

  • void EnableWinograd(bool useWinograd)

    • Summary: Enables or disables the Winograd compute branch. The Winograd compute branch can speed up * 3x3 Convolution at a small loss of accuracy. * useWinograd true to enable the Winograd compute branch. The default is true.
    • Parameter useWinograd: The useWinograd parameter.

  • long GetPerfProfile(IntPtr timings)

    • Summary: Returns overall time for inference and timings (in ticks) for layers. * * Indexes in returned vector correspond to layers ids. Some layers can be fused with others, * in this case zero ticks count will be return for that skipped layers. Supported by DNN_BACKEND_OPENCV on DNN_TARGET_CPU only. * * timings vector for tick timings for all layers. Returns: overall ticks for model inference.
    • Parameter timings: The timings parameter.
    • Returns: The returned value.

  • void EnableKVCache()

    • Summary: Enables KV-Cache for all AttentionOnnxI layers

  • void DisableKVCache()

    • Summary: Disables KV-Cache for all AttentionOnnxI layers

  • void ResetKVCache()

    • Summary: Resets KV-Cache for all AttentionOnnxI layers

  • void GetPerfProfile(IntPtr names, IntPtr timems, IntPtr counts)

    • Summary: Returns profiling data captured during the last forward pass. * * Entries are sorted by time in descending order. Empty vectors are returned * if profiling is disabled (DNN_PROFILE_NONE).
    • Parameter names: The names parameter.
    • Parameter timems: The timems parameter.
    • Parameter counts: The counts parameter.

  • void PrintPerfProfile()

    • Summary: Prints the profile captured during the last forward pass in a formatted table using CV_LOG_INFO. * * In DNN_PROFILE_DETAILED mode, prints per-layer label, time, and percentage. * In DNN_PROFILE_SUMMARY mode, prints per-type count, time, and percentage. * Does nothing if profiling is disabled (DNN_PROFILE_NONE) or all timings are zero.

DnnSegmentationModel ​

Inherits from: DnnModel

This class represents high-level API for segmentation models * * SegmentationModel allows to set params for preprocessing input image. * SegmentationModel creates net from file with trained weights and config, * sets preprocessing input, runs forward pass and returns the class prediction for each pixel.

Constructors ​

  • new DnnSegmentationModel(string model, string? config)
    • Summary: * Create segmentation model from network represented in one of the supported formats.
    • Parameter model: The model parameter.
    • Parameter config: The config parameter.
  • new DnnSegmentationModel(DnnNet network)
    • Summary: * Create model from deep learning network.
    • Parameter network: The network parameter.

Methods ​

  • void Segment(Mat frame, Mat mask)
    • Summary: Given the input frame, create input blob, run net * frame The input image. * mask Allocated class prediction for each pixel
    • Parameter frame: The frame parameter.
    • Parameter mask: Optional operation mask.

DnnTextDetectionModel ​

Inherits from: DnnModel

Base class for text detection networks

Methods ​

  • void Detect(Mat frame, IntPtr detections, IntPtr confidences)
    • Summary: Performs detection * * Given the input frame, prepare network input, run network inference, post-process network output and return result detections. * * Each result is quadrangle's 4 points in this order: * - bottom-left * - top-left * - top-right * - bottom-right * * Use getPerspectiveTransform function to retrieve image region without perspective transformations. * * Note: If DL model doesn't support that kind of output then result may be derived from detectTextRectangles() output. * * frame The input image * detections array with detections' quadrangles (4 points per result) * confidences array with detection confidences
    • Parameter frame: The frame parameter.
    • Parameter detections: The detections parameter.
    • Parameter confidences: The confidences parameter.
  • void Detect(Mat frame, IntPtr detections)
    • Summary: This is an overloaded member function, provided for convenience.
    • Parameter frame: The frame parameter.
    • Parameter detections: The detections parameter.
  • void DetectTextRectangles(Mat frame, IntPtr detections, IntPtr confidences)
    • Summary: Performs detection * * Given the input frame, prepare network input, run network inference, post-process network output and return result detections. * * Each result is rotated rectangle. * * Note: Result may be inaccurate in case of strong perspective transformations. * * frame the input image * detections array with detections' RotationRect results * confidences array with detection confidences
    • Parameter frame: The frame parameter.
    • Parameter detections: The detections parameter.
    • Parameter confidences: The confidences parameter.
  • void DetectTextRectangles(Mat frame, IntPtr detections)
    • Summary: This is an overloaded member function, provided for convenience.
    • Parameter frame: The frame parameter.
    • Parameter detections: The detections parameter.

DnnTextDetectionModelDb ​

Inherits from: DnnTextDetectionModel

This class represents high-level API for text detection DL networks compatible with DB model. * * Related publications: [liao2020real] * Paper: https://arxiv.org/abs/1911.08947 * For more information about the hyper-parameters setting, please refer to https://github.com/MhLiao/DB * * Configurable parameters: * - (float) binaryThreshold - The threshold of the binary map. It is usually set to 0.3. * - (float) polygonThreshold - The threshold of text polygons. It is usually set to 0.5, 0.6, and 0.7. Default is 0.5f * - (double) unclipRatio - The unclip ratio of the detected text region, which determines the output size. It is usually set to 2.0. * - (int) maxCandidates - The max number of the output results.

Constructors ​

  • new DnnTextDetectionModelDb(DnnNet network)
    • Summary: * Create text detection algorithm from deep learning network.
    • Parameter network: The network parameter.
  • new DnnTextDetectionModelDb(string model, string? config)
    • Summary: * Create text detection model from network represented in one of the supported formats.
    • Parameter model: The model parameter.
    • Parameter config: The config parameter.

Methods ​

  • DnnTextDetectionModelDb? SetBinaryThreshold(float binaryThreshold)
    • Summary: Sets the threshold for the binary map. Typically set to 0.3.
    • Parameter binaryThreshold: The binary map threshold value.
    • Returns: The DnnTextDetectionModelDb instance for method chaining, or null.
  • float GetBinaryThreshold()
    • Summary: Gets the current binary map threshold value.
    • Returns: The current binary threshold.
  • DnnTextDetectionModelDb? SetPolygonThreshold(float polygonThreshold)
    • Summary: Sets the threshold for text polygons. Typically set to 0.5, 0.6, or 0.7.
    • Parameter polygonThreshold: The polygon threshold value.
    • Returns: The DnnTextDetectionModelDb instance for method chaining, or null.
  • float GetPolygonThreshold()
    • Summary: Gets the current text polygon threshold value.
    • Returns: The current polygon threshold.
  • DnnTextDetectionModelDb? SetUnclipRatio(double unclipRatio)
    • Summary: Sets the unclip ratio of the detected text region, which determines the output size. Typically set to 2.0.
    • Parameter unclipRatio: The unclip ratio value.
    • Returns: The DnnTextDetectionModelDb instance for method chaining, or null.
  • double GetUnclipRatio()
    • Summary: Gets the current unclip ratio value.
    • Returns: The current unclip ratio.
  • DnnTextDetectionModelDb? SetMaxCandidates(int maxCandidates)
    • Summary: Sets the maximum number of output text detection results.
    • Parameter maxCandidates: The maximum number of candidate results.
    • Returns: The DnnTextDetectionModelDb instance for method chaining, or null.
  • int GetMaxCandidates()
    • Summary: Gets the current maximum number of output text detection results.
    • Returns: The current max candidates value.

DnnTextDetectionModelEast ​

Inherits from: DnnTextDetectionModel

This class represents high-level API for text detection DL networks compatible with EAST model. * * Configurable parameters: * - (float) confThreshold - used to filter boxes by confidences, default: 0.5f * - (float) nmsThreshold - used in non maximum suppression, default: 0.0f

Constructors ​

  • new DnnTextDetectionModelEast(DnnNet network)
    • Summary: * Create text detection algorithm from deep learning network
    • Parameter network: The network parameter.
  • new DnnTextDetectionModelEast(string model, string? config)
    • Summary: * Create text detection model from network represented in one of the supported formats.
    • Parameter model: The model parameter.
    • Parameter config: The config parameter.

Methods ​

  • DnnTextDetectionModelEast? SetConfidenceThreshold(float confThreshold)

    • Summary: * Set the detection confidence threshold
    • Remarks:

    • Parameter confThreshold: A threshold used to filter boxes by confidences

    • Parameter confThreshold: The confThreshold parameter.

    • Returns: The returned value.

  • float GetConfidenceThreshold()

    • Summary: * Get the detection confidence threshold
    • Returns: The returned value.

  • DnnTextDetectionModelEast? SetNMSThreshold(float nmsThreshold)

    • Summary: * Set the detection NMS filter threshold
    • Remarks:

    • Parameter nmsThreshold: A threshold used in non maximum suppression

    • Parameter nmsThreshold: The nmsThreshold parameter.

    • Returns: The returned value.

  • float GetNMSThreshold()

    • Summary: * Get the detection confidence threshold
    • Returns: The returned value.

DnnTextRecognitionModel ​

Inherits from: DnnModel

This class represents high-level API for text recognition networks. * * TextRecognitionModel allows to set params for preprocessing input image. * TextRecognitionModel creates net from file with trained weights and config, * sets preprocessing input, runs forward pass and return recognition result. * For TextRecognitionModel, CRNN-CTC is supported.

Constructors ​

  • new DnnTextRecognitionModel(DnnNet network)
    • Summary: * Create Text Recognition model from deep learning network
    • Parameter network: The network parameter.
  • new DnnTextRecognitionModel(string model, string? config)
    • Summary: * Create text recognition model from network represented in one of the supported formats
    • Parameter model: The model parameter.
    • Parameter config: The config parameter.

Methods ​

  • DnnTextRecognitionModel? SetDecodeType(string decodeType)

    • Summary: * Set the decoding method of translating the network output into string
    • Remarks:
    • Parameter decodeType: The decoding method of translating the network output into string, currently supported type:
    • "CTC-greedy" greedy decoding for the output of CTC-based methods
    • "CTC-prefix-beam-search" Prefix beam search decoding for the output of CTC-based methods

  • Parameter decodeType: The decodeType parameter.

  • Returns: The returned value.

  • string? GetDecodeType()

    • Summary: * Get the decoding method
    • Remarks:

Returns: the decoding method

  • Returns: The returned value.

  • DnnTextRecognitionModel? SetDecodeOptsCTCPrefixBeamSearch(int beamSize, int vocPruneSize)

    • Summary: * Set the decoding method options for "CTC-prefix-beam-search" decode usage
    • Remarks:
    • Parameter beamSize: Beam size for search
    • Parameter vocPruneSize: Parameter to optimize big vocabulary search,

  • only take top vocPruneSize tokens in each search step, vocPruneSize <= 0 stands for disable this prune.

    • Parameter beamSize: The beamSize parameter.
    • Parameter vocPruneSize: The vocPruneSize parameter.
    • Returns: The returned value.

  • DnnTextRecognitionModel? SetVocabulary(IntPtr vocabulary)

    • Summary: * Set the vocabulary for recognition.
    • Remarks:

    • Parameter vocabulary: the associated vocabulary of the network.

    • Parameter vocabulary: The vocabulary parameter.

    • Returns: The returned value.

  • IntPtr GetVocabulary()

    • Summary: * Get the vocabulary for recognition.
    • Remarks:

Returns: vocabulary the associated vocabulary

  • Returns: The returned value.

  • string? Recognize(Mat frame)

    • Summary: * Given the input frame, create input blob, run net and return recognition result
    • Remarks:

    • Parameter frame: The input image Returns: The text recognition result

    • Parameter frame: The frame parameter.

    • Returns: The returned value.

  • void Recognize(Mat frame, IntPtr roiRects, IntPtr results)

    • Summary: * Given the input frame, create input blob, run net and return recognition result
    • Remarks:
    • Parameter frame: The input image
    • Parameter roiRects: List of text detection regions of interest (Rect, CV_32SC4). ROIs is be cropped as the network inputs

    • Parameter results: A set of text recognition results.

    • Parameter frame: The frame parameter.

    • Parameter roiRects: The roiRects parameter.

    • Parameter results: The results parameter.

DnnTokenizer ​

Inherits from: DisposableOpenCVObject

  • High-level tokenizer wrapper for DNN usage.

Detailed Remarks:

  • Provides a simple API to encode and decode tokens for LLMs.

  • Models are loaded via Tokenizer.load().

csharp
using var tok = DnnTokenizer.Load("/path/to/model/");
IntPtr ids = tok.Encode("hello world");
string? text = tok.Decode(ids);

Methods ​

  • DnnTokenizer? Load(string model_config)

    • Summary: * Load a tokenizer from a model directory.
    • Remarks:

  • Expects the directory to contain:

    • config.json with field model_type with value "gpt2" or "gpt4".
    • tokenizer.json produced by the corresponding model family.

  • The argument is a path prefix; this function concatenates file

  • names directly (e.g. model_dir + "config.json"), so model_dir must

  • end with an appropriate path separator.

    • Parameter model_config: Path to config.json for model. Returns: A Tokenizer ready for use. Throws OpenCVException if files are missing or model_type is unsupported.

    • Parameter model_config: The model_config parameter.

    • Returns: The returned value.

  • IntPtr Encode(string text)

    • Summary: * Encode UTF-8 text to token ids (special tokens currently disabled).
    • Remarks:

  • Calls the underlying CoreBPE.encode with an empty allowed-special set.

    • Parameter text: UTF-8 input string. Returns: Vector of token ids (32-bit ids narrowed to int for convenience).

    • Parameter text: The text parameter.

    • Returns: The returned value.

  • string? Decode(IntPtr tokens)

    • Summary: Decodes token ids back to a UTF-8 string.
    • Parameter tokens: Pointer to a vector of token ids to decode.
    • Returns: The decoded UTF-8 string, or null.

⚙️ Static Methods (Cv2) ​

Cv2.DnnGetAvailableTargets ​

Signature: IntPtr DnnGetAvailableTargets(IntPtr be)

Retrieves the list of available hardware targets for a given DNN backend.

Parameters:

  • be: The be parameter.

Returns: The returned value.

Cv2.DnnReadNetFromTensorflow ​

Signature: DnnNet? DnnReadNetFromTensorflow(string model, string? config, int engine, IntPtr extraOutputs)

Reads a network model stored in TensorFlow framework's format. * model path to the .pb file with binary protobuf description of the network architecture * config path to the .pbtxt file that contains text graph definition in protobuf format. * Resulting Net object is built by text graph using weights from a binary one that * let us make it more flexible. * engine select DNN engine to be used. With auto selection the new engine is used. * extraOutputs specify model outputs explicitly, in addition to the outputs the graph analyzer finds. * Please pay attention that the new DNN does not support non-CPU back-ends for now. * Returns: Net object.

Parameters:

  • model: The model parameter.
  • config: The config parameter.
  • engine: The engine parameter.
  • extraOutputs: The extraOutputs parameter.

Returns: The returned value.

Cv2.DnnReadNetFromTensorflow ​

Signature: DnnNet? DnnReadNetFromTensorflow(IntPtr bufferModel, IntPtr bufferConfig, int engine, IntPtr extraOutputs)

Reads a network model stored in TensorFlow framework's format. * bufferModel buffer containing the content of the pb file * bufferConfig buffer containing the content of the pbtxt file * engine select DNN engine to be used. With auto selection the new engine is used. * extraOutputs specify model outputs explicitly, in addition to the outputs the graph analyzer finds. * Please pay attention that the new DNN does not support non-CPU back-ends for now. * Returns: Net object.

Parameters:

  • bufferModel: The bufferModel parameter.
  • bufferConfig: The bufferConfig parameter.
  • engine: The engine parameter.
  • extraOutputs: The extraOutputs parameter.

Returns: The returned value.

Cv2.DnnReadNetFromTFLite ​

Signature: DnnNet? DnnReadNetFromTFLite(string model, int engine)

Reads a network model stored in TFLite framework's format. * model path to the .tflite file with binary flatbuffers description of the network architecture * engine select DNN engine to be used. With auto selection the new engine is used first and falls back to classic. * Please pay attention that the new DNN does not support non-CPU back-ends for now. * Returns: Net object.

Parameters:

  • model: The model parameter.
  • engine: The engine parameter.

Returns: The returned value.

Cv2.DnnReadNetFromTFLite ​

Signature: DnnNet? DnnReadNetFromTFLite(IntPtr bufferModel, int engine)

Reads a network model stored in TFLite framework's format. * bufferModel buffer containing the content of the tflite file * engine select DNN engine to be used. With auto selection the new engine is used first and falls back to classic. * Please pay attention that the new DNN does not support non-CPU back-ends for now. * Returns: Net object.

Parameters:

  • bufferModel: The bufferModel parameter.
  • engine: The engine parameter.

Returns: The returned value.

Cv2.DnnReadNet ​

Signature: DnnNet? DnnReadNet(string model, string? config, string? framework, int engine)

  • Read deep learning network represented in one of the supported formats.

Detailed Remarks:

    • Parameter model: Binary file contains trained weights. The following file
  •              extensions are expected for models from different frameworks:

  •              * `*.pb` (TensorFlow, https://www.tensorflow.org/)

  •              * `*.bin` | `*.onnx` (OpenVINO, https://software.intel.com/openvino-toolkit)

  •              * `*.onnx` (ONNX, https://onnx.ai/)
    • Parameter config: Text file contains network configuration. It could be a
  •               file with the following extensions:

  •              * `*.pbtxt` (TensorFlow, https://www.tensorflow.org/)

  •              * `*.xml` (OpenVINO, https://software.intel.com/openvino-toolkit)
    • Parameter framework: Explicit framework name tag to determine a format.
    • Parameter engine: select DNN engine to be used. With auto selection the new engine is used first and falls back to classic.

  • Please pay attention that the new DNN does not support non-CPU back-ends for now.

  • Use ENGINE_CLASSIC if you want to use other back-ends.

  • Returns: Net object.

  • This function automatically detects an origin framework of trained model

  • and calls an appropriate function such readNetFromTensorflow, readNetFromONNX.

  • An order of model and config arguments does not matter.

Parameters:

  • model: The model parameter.
  • config: The config parameter.
  • framework: The framework parameter.
  • engine: The engine parameter.

Returns: The returned value.

Cv2.DnnReadNet ​

Signature: DnnNet? DnnReadNet(string framework, IntPtr bufferModel, IntPtr bufferConfig, int engine)

  • Read deep learning network represented in one of the supported formats.

Detailed Remarks:

  • This is an overloaded member function, provided for convenience.
  •      It differs from the above function only in what argument(s) it accepts.
    • Parameter framework: Name of origin framework.
    • Parameter bufferModel: A buffer with a content of binary file with weights
    • Parameter bufferConfig: A buffer with a content of text file contains network configuration.
    • Parameter engine: select DNN engine to be used. With auto selection the new engine is used first and falls back to classic.
  • Please pay attention that the new DNN does not support non-CPU back-ends for now.
  • Use ENGINE_CLASSIC if you want to use other back-ends.
  • Returns: Net object.

Parameters:

  • framework: The framework parameter.
  • bufferModel: The bufferModel parameter.
  • bufferConfig: The bufferConfig parameter.
  • engine: The engine parameter.

Returns: The returned value.

Cv2.DnnReadNetFromModelOptimizer ​

Signature: DnnNet? DnnReadNetFromModelOptimizer(string xml, string? bin)

Load a network from Intel's Model Optimizer intermediate representation. * xml XML configuration file with network's topology. * bin Binary file with trained weights. * Returns: Net object. * Networks imported from Intel's Model Optimizer are launched in Intel's Inference Engine * backend.

Parameters:

  • xml: The xml parameter.
  • bin: The bin parameter.

Returns: The returned value.

Cv2.DnnReadNetFromModelOptimizer ​

Signature: DnnNet? DnnReadNetFromModelOptimizer(IntPtr bufferModelConfig, IntPtr bufferWeights)

Load a network from Intel's Model Optimizer intermediate representation. * bufferModelConfig Buffer contains XML configuration with network's topology. * bufferWeights Buffer contains binary data with trained weights. * Returns: Net object. * Networks imported from Intel's Model Optimizer are launched in Intel's Inference Engine * backend.

Parameters:

  • bufferModelConfig: The bufferModelConfig parameter.
  • bufferWeights: The bufferWeights parameter.

Returns: The returned value.

Cv2.DnnReadNetFromONNX ​

Signature: DnnNet? DnnReadNetFromONNX(string onnxFile, int engine)

Reads a network model ONNX. * onnxFile path to the .onnx file with text description of the network architecture. * engine select DNN engine to be used. With auto selection the new engine is used first and falls back to classic. * Please pay attention that the new DNN does not support non-CPU back-ends for now. * Returns: Network object that ready to do forward, throw an exception in failure cases.

Parameters:

  • onnxFile: The onnxFile parameter.
  • engine: The engine parameter.

Returns: The returned value.

Cv2.DnnReadNetFromONNX ​

Signature: DnnNet? DnnReadNetFromONNX(IntPtr buffer, int engine)

Reads a network model from ONNX * in-memory buffer. * buffer in-memory buffer that stores the ONNX model bytes. * engine select DNN engine to be used. With auto selection the new engine is used first and falls back to classic. * Please pay attention that the new DNN does not support non-CPU back-ends for now. * Returns: Network object that ready to do forward, throw an exception * in failure cases.

Parameters:

  • buffer: The buffer parameter.
  • engine: The engine parameter.

Returns: The returned value.

Cv2.DnnReadTensorFromONNX ​

Signature: Mat? DnnReadTensorFromONNX(string path)

Creates blob from .pb file. * path to the .pb file with input tensor. * Returns: Mat.

Parameters:

  • path: The path parameter.

Returns: The returned value.

Cv2.DnnBlobFromImage ​

Signature: Mat? DnnBlobFromImage(Mat image, double scalefactor, Size size, Scalar mean, bool swapRB, bool crop, int ddepth)

Creates 4-dimensional blob from image. Optionally resizes and crops image from center, * subtract mean values, scales values by scalefactor, swap Blue and Red channels. * image input image (with 1-, 3- or 4-channels). * scalefactor multiplier for images values. * size spatial size for output image * mean scalar with mean values which are subtracted from channels. Values are intended * to be in (mean-R, mean-G, mean-B) order if image has BGR ordering and swapRB is true. * swapRB flag which indicates that swap first and last channels * in 3-channel image is necessary. * crop flag which indicates whether image will be cropped after resize or not * ddepth Depth of output blob. Choose CV_32F or CV_8U. * if crop is true, input image is resized so one side after resize is equal to corresponding * dimension in size and another one is equal or larger. Then, crop from the center is performed. * If crop is false, direct resize without cropping and preserving aspect ratio is performed. * Returns: 4-dimensional Mat with NCHW dimensions order. * * Note: * The order and usage of scalefactor and mean are (input - mean) * scalefactor.

Parameters:

  • image: Input image.
  • scalefactor: The scalefactor parameter.
  • size: The size parameter.
  • mean: The mean parameter.
  • swapRB: The swapRB parameter.
  • crop: The crop parameter.
  • ddepth: The ddepth parameter.

Returns: The returned value.

Cv2.DnnBlobFromImages ​

Signature: Mat? DnnBlobFromImages(IntPtr images, double scalefactor, Size size, Scalar mean, bool swapRB, bool crop, int ddepth)

Creates 4-dimensional blob from series of images. Optionally resizes and * crops images from center, subtract mean values, scales values by scalefactor, * swap Blue and Red channels. * images input images (all with 1-, 3- or 4-channels). * size spatial size for output image * mean scalar with mean values which are subtracted from channels. Values are intended * to be in (mean-R, mean-G, mean-B) order if image has BGR ordering and swapRB is true. * scalefactor multiplier for images values. * swapRB flag which indicates that swap first and last channels * in 3-channel image is necessary. * crop flag which indicates whether image will be cropped after resize or not * ddepth Depth of output blob. Choose CV_32F or CV_8U. * if crop is true, input image is resized so one side after resize is equal to corresponding * dimension in size and another one is equal or larger. Then, crop from the center is performed. * If crop is false, direct resize without cropping and preserving aspect ratio is performed. * Returns: 4-dimensional Mat with NCHW dimensions order. * * Note: * The order and usage of scalefactor and mean are (input - mean) * scalefactor.

Parameters:

  • images: The images parameter.
  • scalefactor: The scalefactor parameter.
  • size: The size parameter.
  • mean: The mean parameter.
  • swapRB: The swapRB parameter.
  • crop: The crop parameter.
  • ddepth: The ddepth parameter.

Returns: The returned value.

Cv2.DnnBlobFromImageWithParams ​

Signature: Mat? DnnBlobFromImageWithParams(Mat image, DnnImage2BlobParams? param)

Creates 4-dimensional blob from image with given params. * * This function is an extension of blobFromImage to meet more image preprocess needs. * Given input image and preprocessing parameters, and function outputs the blob. * * image input image (all with 1-, 3- or 4-channels). * param struct of Image2BlobParams, contains all parameters needed by processing of image to blob. Returns: 4-dimensional Mat.

Parameters:

  • image: Input image.
  • param: The param parameter.

Returns: The returned value.

Cv2.DnnBlobFromImageWithParams ​

Signature: void DnnBlobFromImageWithParams(Mat image, Mat blob, DnnImage2BlobParams? param)

This is an overloaded member function, provided for convenience.

Parameters:

  • image: Input image.
  • blob: The blob parameter.
  • param: The param parameter.

Cv2.DnnBlobFromImagesWithParams ​

Signature: Mat? DnnBlobFromImagesWithParams(IntPtr images, DnnImage2BlobParams? param)

Creates 4-dimensional blob from series of images with given params. * * This function is an extension of blobFromImages to meet more image preprocess needs. * Given input image and preprocessing parameters, and function outputs the blob. * * images input image (all with 1-, 3- or 4-channels). * param struct of Image2BlobParams, contains all parameters needed by processing of image to blob. * Returns: 4-dimensional Mat.

Parameters:

  • images: The images parameter.
  • param: The param parameter.

Returns: The returned value.

Cv2.DnnBlobFromImagesWithParams ​

Signature: void DnnBlobFromImagesWithParams(IntPtr images, Mat blob, DnnImage2BlobParams? param)

This is an overloaded member function, provided for convenience.

Parameters:

  • images: The images parameter.
  • blob: The blob parameter.
  • param: The param parameter.

Cv2.DnnImagesFromBlob ​

Signature: void DnnImagesFromBlob(Mat blob_, IntPtr images_)

Parse a 4D blob and output the images it contains as 2D arrays through a simpler data structure * (Mat[]). * blob_ 4 dimensional array (images, channels, height, width) in floating point precision (CV_32F) from * which you would like to extract the images. * images_ array of 2D Mat containing the images extracted from the blob in floating point precision * (CV_32F). They are non normalized neither mean added. The number of returned images equals the first dimension * of the blob (batch size). Every image has a number of channels equals to the second dimension of the blob (depth).

Parameters:

  • blob_: The blob_ parameter.
  • images_: The images_ parameter.

Cv2.DnnWriteTextGraph ​

Signature: void DnnWriteTextGraph(string model, string output)

Create a text representation for a binary network stored in protocol buffer format. * model A path to binary network. * output A path to output text file to be created. * * Note: To reduce output file size, trained weights are not included.

Parameters:

  • model: The model parameter.
  • output: The output parameter.

Cv2.DnnNMSBoxes ​

Signature: void DnnNMSBoxes(IntPtr bboxes, IntPtr scores, float score_threshold, float nms_threshold, IntPtr indices, float eta, int top_k)

Performs non maximum suppression given boxes and corresponding scores.

Detailed Remarks:

    • Parameter bboxes: a set of bounding boxes to apply NMS.
    • Parameter scores: a set of corresponding confidences.
    • Parameter score_threshold: a threshold used to filter boxes by score.
    • Parameter nms_threshold: a threshold used in non maximum suppression.
    • Parameter indices: the kept indices of bboxes after NMS.
    • Parameter eta: a coefficient in adaptive threshold formula: formula.
    • Parameter top_k: if >0, keep at most top_k picked indices.

Parameters:

  • bboxes: The bboxes parameter.
  • scores: The scores parameter.
  • score_threshold: The score_threshold parameter.
  • nms_threshold: The nms_threshold parameter.
  • indices: The indices parameter.
  • eta: The eta parameter.
  • top_k: The top_k parameter.

Cv2.DnnNMSBoxesBatched ​

Signature: void DnnNMSBoxesBatched(IntPtr bboxes, IntPtr scores, IntPtr class_ids, float score_threshold, float nms_threshold, IntPtr indices, float eta, int top_k)

Performs batched non maximum suppression on given boxes and corresponding scores across different classes.

Detailed Remarks:

    • Parameter bboxes: a set of bounding boxes to apply NMS.
    • Parameter scores: a set of corresponding confidences.
    • Parameter class_ids: a set of corresponding class ids. Ids are integer and usually start from 0.
    • Parameter score_threshold: a threshold used to filter boxes by score.
    • Parameter nms_threshold: a threshold used in non maximum suppression.
    • Parameter indices: the kept indices of bboxes after NMS.
    • Parameter eta: a coefficient in adaptive threshold formula: formula.
    • Parameter top_k: if >0, keep at most top_k picked indices.

Parameters:

  • bboxes: The bboxes parameter.
  • scores: The scores parameter.
  • class_ids: The class_ids parameter.
  • score_threshold: The score_threshold parameter.
  • nms_threshold: The nms_threshold parameter.
  • indices: The indices parameter.
  • eta: The eta parameter.
  • top_k: The top_k parameter.

Cv2.DnnSoftNMSBoxes ​

Signature: void DnnSoftNMSBoxes(IntPtr bboxes, IntPtr scores, IntPtr updated_scores, float score_threshold, float nms_threshold, IntPtr indices, long top_k, float sigma, DnnSoftNMSMethod method)

Performs soft non maximum suppression given boxes and corresponding scores. * Reference: https://arxiv.org/abs/1704.04503 * bboxes a set of bounding boxes to apply Soft NMS. * scores a set of corresponding confidences. * updated_scores a set of corresponding updated confidences. * score_threshold a threshold used to filter boxes by score. * nms_threshold a threshold used in non maximum suppression. * indices the kept indices of bboxes after NMS. * top_k keep at most top_k picked indices. * sigma parameter of Gaussian weighting. * method Gaussian or linear. * See also: SoftNMSMethod

Parameters:

  • bboxes: The bboxes parameter.
  • scores: The scores parameter.
  • updated_scores: The updated_scores parameter.
  • score_threshold: The score_threshold parameter.
  • nms_threshold: The nms_threshold parameter.
  • indices: The indices parameter.
  • top_k: The top_k parameter.
  • sigma: The sigma parameter.
  • method: The method parameter.

Cv2.DnnGetInferenceEngineBackendType ​

Signature: string? DnnGetInferenceEngineBackendType()

Returns Inference Engine internal backend API. * * See values of CV_DNN_BACKEND_INFERENCE_ENGINE_* macros. * * OPENCV_DNN_BACKEND_INFERENCE_ENGINE_TYPE runtime parameter (environment variable) is ignored since 4.6.0. * * (Deprecated)

Returns: The returned value.

Cv2.DnnSetInferenceEngineBackendType ​

Signature: string? DnnSetInferenceEngineBackendType(string newBackendType)

Specify Inference Engine internal backend API. * * See values of CV_DNN_BACKEND_INFERENCE_ENGINE_* macros. * * Returns: previous value of internal backend API * * (Deprecated)

Parameters:

  • newBackendType: The newBackendType parameter.

Returns: The returned value.

Cv2.DnnResetMyriadDevice ​

Signature: void DnnResetMyriadDevice()

Release a Myriad device (binded by OpenCV). * * Single Myriad device cannot be shared across multiple processes which uses * Inference Engine's Myriad plugin.

Cv2.DnnGetInferenceEngineVPUType ​

Signature: string? DnnGetInferenceEngineVPUType()

Returns Inference Engine VPU type. * * See values of CV_DNN_INFERENCE_ENGINE_VPU_TYPE_* macros.

Returns: The returned value.

Cv2.DnnGetInferenceEngineCPUType ​

Signature: string? DnnGetInferenceEngineCPUType()

Returns Inference Engine CPU type. * * Specify OpenVINO plugin: CPU or ARM.

Returns: The returned value.

Cv2.DnnReleaseHDDLPlugin ​

Signature: void DnnReleaseHDDLPlugin()

Release a HDDL plugin.

🔢 Enumerations ​

DnnActivationType ​

Wrapper for OpenCV's native functionality.

MemberValueDescription
None0None
Mishunchecked((int)(0 + 1))Mish
Swishunchecked((int)(0 + 2))Swish
Sigmoidunchecked((int)(0 + 3))Sigmoid
Tanhunchecked((int)(0 + 4))Tanh
Eluunchecked((int)(0 + 5))Elu
Hardswishunchecked((int)(0 + 6))Hardswish
Hardsigmoidunchecked((int)(0 + 7))Hardsigmoid
Geluunchecked((int)(0 + 8))Gelu
GeluApproxunchecked((int)(0 + 9))GeluApprox
Reluunchecked((int)(0 + 10))Relu
Clipunchecked((int)(0 + 11))Clip

DnnArgKind ​

Wrapper for OpenCV's native functionality.

MemberValueDescription
Empty0Empty
Const1Const
Input2Input
Output3Output
Temp4Temp
Pattern5Pattern

DnnAutoPadding ​

Wrapper for OpenCV's native functionality.

MemberValueDescription
None0None
SameUpper1SameUpper
SameLower2SameLower
Valid3Valid

DnnBackend ​

Wrapper for OpenCV's native functionality.

MemberValueDescription
Default0Default
InferenceEngine2InferenceEngine
Opencvunchecked((int)(2 + 1))Opencv
Vkcomunchecked((int)(2 + 2))Vkcom
Cudaunchecked((int)(2 + 3))Cuda
Webnnunchecked((int)(2 + 4))Webnn
Timvxunchecked((int)(2 + 5))Timvx
Cannunchecked((int)(2 + 6))Cann

DnnEngineType ​

Wrapper for OpenCV's native functionality.

MemberValueDescription
Classic1Classic
New2New
Auto3Auto
Ort4Ort

DnnImagePaddingMode ​

Wrapper for OpenCV's native functionality.

MemberValueDescription
Null0Null
CropCenter1CropCenter
Letterbox2Letterbox

DnnLossReduction ​

Wrapper for OpenCV's native functionality.

MemberValueDescription
None0None
Mean1Mean
Sum2Sum

DnnModelFormat ​

Wrapper for OpenCV's native functionality.

MemberValueDescription
Generic0Generic
Onnx1Onnx
Tf2Tf
Tflite3Tflite

DnnNaryEltwiseLayerOperation ​

Wrapper for OpenCV's native functionality.

MemberValueDescription
And0And
Equalunchecked((int)(0 + 1))Equal
Greaterunchecked((int)(0 + 2))Greater
GreaterEqualunchecked((int)(0 + 3))GreaterEqual
Lessunchecked((int)(0 + 4))Less
LessEqualunchecked((int)(0 + 5))LessEqual
Orunchecked((int)(0 + 6))Or
Powunchecked((int)(0 + 7))Pow
Xorunchecked((int)(0 + 8))Xor
Bitshiftunchecked((int)(0 + 9))Bitshift
Maxunchecked((int)(0 + 10))Max
Meanunchecked((int)(0 + 11))Mean
Minunchecked((int)(0 + 12))Min
Modunchecked((int)(0 + 13))Mod
Fmodunchecked((int)(0 + 14))Fmod
Produnchecked((int)(0 + 15))Prod
Subunchecked((int)(0 + 16))Sub
Sumunchecked((int)(0 + 17))Sum
Addunchecked((int)(0 + 18))Add
Divunchecked((int)(0 + 19))Div
Whereunchecked((int)(0 + 20))Where
BitwiseAndunchecked((int)(0 + 21))BitwiseAnd
BitwiseOrunchecked((int)(0 + 22))BitwiseOr
BitwiseXorunchecked((int)(0 + 23))BitwiseXor

DnnProfilingMode ​

Wrapper for OpenCV's native functionality.

MemberValueDescription
None0None
Summary1Summary
Detailed2Detailed

DnnReduce2LayerReduceType ​

Wrapper for OpenCV's native functionality.

MemberValueDescription
Max0Max
Min1Min
Mean2Mean
Sum3Sum
L14L1
L25L2
Prod6Prod
SumSquare7SumSquare
LogSum8LogSum
LogSumExp9LogSumExp

DnnSoftNMSMethod ​

Wrapper for OpenCV's native functionality.

MemberValueDescription
Linear1Linear
Gaussian2Gaussian

DnnTarget ​

Wrapper for OpenCV's native functionality.

MemberValueDescription
Cpu0Cpu
Openclunchecked((int)(0 + 1))Opencl
OpenclFp16unchecked((int)(0 + 2))OpenclFp16
Myriadunchecked((int)(0 + 3))Myriad
Vulkanunchecked((int)(0 + 4))Vulkan
Fpgaunchecked((int)(0 + 5))Fpga
Cudaunchecked((int)(0 + 6))Cuda
CudaFp16unchecked((int)(0 + 7))CudaFp16
Hddlunchecked((int)(0 + 8))Hddl
Npuunchecked((int)(0 + 9))Npu
CpuFp16unchecked((int)(0 + 10))CpuFp16

DnnTracingMode ​

Wrapper for OpenCV's native functionality.

MemberValueDescription
None0None
All1All
Op2Op

Released under the Apache 2.0 and LGPL 2.1 Licenses.

Copyright © 2026 Qourex