Struct Net

pub struct Net { /* private fields */ }

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.

Implementations

impl Net

pub fn default() -> Result<Net>

pub fn read_from_model_optimizer(xml: &str, bin: &str) -> Result<Net>

Create a network from Intel’s Model Optimizer intermediate representation (IR).

Parameters

• 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.

pub fn read_from_model_optimizer_1( buffer_model_config: &Vector<u8>, buffer_weights: &Vector<u8>, ) -> Result<Net>

Create a network from Intel’s Model Optimizer in-memory buffers with intermediate representation (IR).

Parameters

• bufferModelConfig: buffer with model’s configuration.
• bufferWeights: buffer with model’s trained weights.

Returns

Net object.

pub fn read_from_model_optimizer_2( buffer_model_config_ptr: &[u8], buffer_weights_ptr: &[u8], ) -> Result<Net>

Create a network from Intel’s Model Optimizer in-memory buffers with intermediate representation (IR).

Parameters

• bufferModelConfigPtr: buffer pointer of model’s configuration.
• bufferModelConfigSize: buffer size of model’s configuration.
• bufferWeightsPtr: buffer pointer of model’s trained weights.
• bufferWeightsSize: buffer size of model’s trained weights.

Returns

Net object.

Trait Implementations

impl Boxed for Net

unsafe fn from_raw(ptr: <Net as OpenCVFromExtern>::ExternReceive) -> Self

Wrap the specified raw pointer

fn into_raw(self) -> <Net as OpenCVTypeExternContainer>::ExternSendMut

Return the underlying raw pointer while consuming this wrapper.

fn as_raw(&self) -> <Net as OpenCVTypeExternContainer>::ExternSend

Return the underlying raw pointer.

fn as_raw_mut(&mut self) -> <Net as OpenCVTypeExternContainer>::ExternSendMut

Return the underlying mutable raw pointer

impl Clone for Net

fn clone(&self) -> Self

Returns a duplicate of the value.

const fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl Debug for Net

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.

impl Drop for Net

fn drop(&mut self)

Executes the destructor for this type.

impl NetTrait for Net

fn as_raw_mut_Net(&mut self) -> *mut c_void

fn dump(&mut self) -> Result<String>

Dump net to String

fn dump_to_file(&mut self, path: &str) -> Result<()>

Dump net structure, hyperparameters, backend, target and fusion to dot file

fn dump_to_pbtxt(&mut self, path: &str) -> Result<()>

Dump net structure, hyperparameters, backend, target and fusion to pbtxt file

fn add_layer_type( &mut self, name: &str, typ: &str, dtype: &i32, params: &mut impl LayerParamsTrait, ) -> Result<i32>

Adds new layer to the net.

fn add_layer( &mut self, name: &str, typ: &str, params: &mut impl LayerParamsTrait, ) -> Result<i32>

Adds new layer to the net.

fn add_layer_to_prev_type( &mut self, name: &str, typ: &str, dtype: &i32, params: &mut impl LayerParamsTrait, ) -> Result<i32>

Adds new layer and connects its first input to the first output of previously added layer.

fn add_layer_to_prev( &mut self, name: &str, typ: &str, params: &mut impl LayerParamsTrait, ) -> Result<i32>

Adds new layer and connects its first input to the first output of previously added layer.

fn connect_first_second(&mut self, out_pin: &str, inp_pin: &str) -> Result<()>

Connects output of the first layer to input of the second layer.

fn connect( &mut self, out_layer_id: i32, out_num: i32, inp_layer_id: i32, inp_num: i32, ) -> Result<()>

Connects #@p outNum output of the first layer to #@p inNum input of the second layer.

fn register_output( &mut self, output_name: &str, layer_id: i32, output_port: i32, ) -> Result<i32>

Registers network output with name

fn set_inputs_names(&mut self, input_blob_names: &Vector<String>) -> Result<()>

Sets outputs names of the network input pseudo layer.

fn set_input_shape(&mut self, input_name: &str, shape: &MatShape) -> Result<()>

Specify shape of network input.

fn forward_single(&mut self, output_name: &str) -> Result<Mat>

Runs forward pass to compute output of layer with name @p outputName.

fn forward_single_def(&mut self) -> Result<Mat>

Runs forward pass to compute output of layer with name @p outputName.

fn forward_async(&mut self, output_name: &str) -> Result<AsyncArray>

Runs forward pass to compute output of layer with name @p outputName.

fn forward_async_def(&mut self) -> Result<AsyncArray>

Runs forward pass to compute output of layer with name @p outputName.

fn forward_layer( &mut self, output_blobs: &mut impl ToOutputArray, output_name: &str, ) -> Result<()>

Runs forward pass to compute output of layer with name @p outputName.

fn forward_layer_def( &mut self, output_blobs: &mut impl ToOutputArray, ) -> Result<()>

Runs forward pass to compute output of layer with name @p outputName.

fn forward( &mut self, output_blobs: &mut impl ToOutputArray, out_blob_names: &Vector<String>, ) -> Result<()>

Runs forward pass to compute outputs of layers listed in @p outBlobNames.

fn forward_and_retrieve( &mut self, output_blobs: &mut Vector<Vector<Mat>>, out_blob_names: &Vector<String>, ) -> Result<()>

Runs forward pass to compute outputs of layers listed in @p outBlobNames.

fn quantize( &mut self, calib_data: &impl ToInputArray, inputs_dtype: i32, outputs_dtype: i32, per_channel: bool, ) -> Result<Net>

Returns a quantized Net from a floating-point Net.

fn quantize_def( &mut self, calib_data: &impl ToInputArray, inputs_dtype: i32, outputs_dtype: i32, ) -> Result<Net>

Returns a quantized Net from a floating-point Net.

fn set_halide_scheduler(&mut self, scheduler: &str) -> Result<()>

Compile Halide layers.

fn set_preferable_backend(&mut self, backend_id: i32) -> Result<()>

Ask network to use specific computation backend where it supported.

fn set_preferable_target(&mut self, target_id: i32) -> Result<()>

Ask network to make computations on specific target device.

fn set_input( &mut self, blob: &impl ToInputArray, name: &str, scalefactor: f64, mean: Scalar, ) -> Result<()>

Sets the new input value for the network

fn set_input_def(&mut self, blob: &impl ToInputArray) -> Result<()>

Sets the new input value for the network

fn set_param( &mut self, layer: i32, num_param: i32, blob: &impl MatTraitConst, ) -> Result<()>

Sets the new value for the learned param of the layer.

fn set_param_1( &mut self, layer_name: &str, num_param: i32, blob: &impl MatTraitConst, ) -> Result<()>

fn enable_fusion(&mut self, fusion: bool) -> Result<()>

Enables or disables layer fusion in the network.

fn enable_winograd(&mut self, use_winograd: bool) -> Result<()>

Enables or disables the Winograd compute branch. The Winograd compute branch can speed up 3x3 Convolution at a small loss of accuracy.

fn get_perf_profile(&mut self, timings: &mut Vector<f64>) -> Result<i64>

Returns overall time for inference and timings (in ticks) for layers.

impl NetTraitConst for Net

fn as_raw_Net(&self) -> *const c_void

fn empty(&self) -> Result<bool>

Returns true if there are no layers in the network.

fn get_layer_id(&self, layer: &str) -> Result<i32>

Converts string name of the layer to the integer identifier.

fn get_layer_names(&self) -> Result<Vector<String>>

fn get_layer(&self, layer_id: i32) -> Result<Ptr<Layer>>

Returns pointer to layer with specified id or name which the network use.

fn get_layer_1(&self, layer_name: &str) -> Result<Ptr<Layer>>

👎Deprecated: Use int getLayerId(const String &layer)

Returns pointer to layer with specified id or name which the network use.

fn get_layer_2(&self, layer_id: &impl DictValueTraitConst) -> Result<Ptr<Layer>>

👎Deprecated: to be removed

Returns pointer to layer with specified id or name which the network use.

fn get_layer_inputs(&self, layer_id: i32) -> Result<Vector<Ptr<Layer>>>

Returns pointers to input layers of specific layer.

fn get_input_details( &self, scales: &mut Vector<f32>, zeropoints: &mut Vector<i32>, ) -> Result<()>

Returns input scale and zeropoint for a quantized Net.

fn get_output_details( &self, scales: &mut Vector<f32>, zeropoints: &mut Vector<i32>, ) -> Result<()>

Returns output scale and zeropoint for a quantized Net.

fn get_param(&self, layer: i32, num_param: i32) -> Result<Mat>

Returns parameter blob of the layer.

fn get_param_def(&self, layer: i32) -> Result<Mat>

Returns parameter blob of the layer.

fn get_param_1(&self, layer_name: &str, num_param: i32) -> Result<Mat>

C++ default parameters

fn get_param_def_1(&self, layer_name: &str) -> Result<Mat>

Note

fn get_unconnected_out_layers(&self) -> Result<Vector<i32>>

Returns indexes of layers with unconnected outputs.

fn get_unconnected_out_layers_names(&self) -> Result<Vector<String>>

Returns names of layers with unconnected outputs.

fn get_layers_shapes( &self, net_input_shapes: &Vector<MatShape>, layers_ids: &mut Vector<i32>, in_layers_shapes: &mut Vector<Vector<MatShape>>, out_layers_shapes: &mut Vector<Vector<MatShape>>, ) -> Result<()>

Returns input and output shapes for all layers in loaded model; preliminary inferencing isn’t necessary.

fn get_layers_shapes_1( &self, net_input_shape: &MatShape, layers_ids: &mut Vector<i32>, in_layers_shapes: &mut Vector<Vector<MatShape>>, out_layers_shapes: &mut Vector<Vector<MatShape>>, ) -> Result<()>

Returns input and output shapes for all layers in loaded model; preliminary inferencing isn’t necessary.

fn get_layer_shapes( &self, net_input_shape: &MatShape, layer_id: i32, in_layer_shapes: &mut Vector<MatShape>, out_layer_shapes: &mut Vector<MatShape>, ) -> Result<()>

Returns input and output shapes for layer with specified id in loaded model; preliminary inferencing isn’t necessary.

fn get_layer_shapes_1( &self, net_input_shapes: &Vector<MatShape>, layer_id: i32, in_layer_shapes: &mut Vector<MatShape>, out_layer_shapes: &mut Vector<MatShape>, ) -> Result<()>

Returns input and output shapes for layer with specified id in loaded model; preliminary inferencing isn’t necessary.

fn get_flops(&self, net_input_shapes: &Vector<MatShape>) -> Result<i64>

Computes FLOP for whole loaded model with specified input shapes.

fn get_flops_1(&self, net_input_shape: &MatShape) -> Result<i64>

Computes FLOP for whole loaded model with specified input shapes.

fn get_flops_2( &self, layer_id: i32, net_input_shapes: &Vector<MatShape>, ) -> Result<i64>

Computes FLOP for whole loaded model with specified input shapes.

fn get_flops_3(&self, layer_id: i32, net_input_shape: &MatShape) -> Result<i64>

Computes FLOP for whole loaded model with specified input shapes.

fn get_layer_types(&self, layers_types: &mut Vector<String>) -> Result<()>

Returns list of types for layer used in model.

fn get_layers_count(&self, layer_type: &str) -> Result<i32>

Returns count of layers of specified type.

fn get_memory_consumption( &self, net_input_shapes: &Vector<MatShape>, weights: &mut size_t, blobs: &mut size_t, ) -> Result<()>

Computes bytes number which are required to store all weights and intermediate blobs for model.

fn get_memory_consumption_1( &self, net_input_shape: &MatShape, weights: &mut size_t, blobs: &mut size_t, ) -> Result<()>

Computes bytes number which are required to store all weights and intermediate blobs for model.

fn get_memory_consumption_for_layer( &self, layer_id: i32, net_input_shapes: &Vector<MatShape>, weights: &mut size_t, blobs: &mut size_t, ) -> Result<()>

Computes bytes number which are required to store all weights and intermediate blobs for model.

fn get_memory_consumption_2( &self, layer_id: i32, net_input_shape: &MatShape, weights: &mut size_t, blobs: &mut size_t, ) -> Result<()>

Computes bytes number which are required to store all weights and intermediate blobs for model.

fn get_memory_consumption_for_layers( &self, net_input_shapes: &Vector<MatShape>, layer_ids: &mut Vector<i32>, weights: &mut Vector<size_t>, blobs: &mut Vector<size_t>, ) -> Result<()>

Computes bytes number which are required to store all weights and intermediate blobs for each layer.

fn get_memory_consumption_3( &self, net_input_shape: &MatShape, layer_ids: &mut Vector<i32>, weights: &mut Vector<size_t>, blobs: &mut Vector<size_t>, ) -> Result<()>

Computes bytes number which are required to store all weights and intermediate blobs for each layer.

impl Send for Net