fast::NeuralNetworkNode class
#include <FAST/Algorithms/NeuralNetwork/InferenceEngine.hpp>

A struct/class representing a neural network input/output node.

Contents

Constructors, destructors, conversion operators

NeuralNetworkNode(std::string name, NodeType type = NodeType::UNSPECIFIED, fast::TensorShape shape = fast::TensorShape(), uint id = 0, fast::TensorShape minShape = fast::TensorShape(), fast::TensorShape maxShape = fast::TensorShape(), fast::TensorShape optShape = fast::TensorShape())
Constructor.
NeuralNetworkNode()

Public variables

uint id
std::string name
NodeType type
fast::TensorShape shape
fast::TensorShape optShape
fast::TensorShape minShape
fast::TensorShape maxShape
std::shared_ptr<fast::Tensor> data

Function documentation

fast::NeuralNetworkNode::NeuralNetworkNode(std::string name, NodeType type = NodeType::UNSPECIFIED, fast::TensorShape shape = fast::TensorShape(), uint id = 0, fast::TensorShape minShape = fast::TensorShape(), fast::TensorShape maxShape = fast::TensorShape(), fast::TensorShape optShape = fast::TensorShape())

Constructor.

Parameters
name Name of node, must match name of node in network file
type
shape
id Input/Output port id
minShape If node has dynamic shape, you can set the minimum shape for optimizations. Note that this is required for the TensorRT engine.
maxShape If node has dynamic shape, you can set the maximum shape for optimizations. Note that this is required for the TensorRT engine.
optShape