AccelOnnxConverter Class

Class for converting models to AccelOnnx flavor.

Inheritance

builtins.object

AccelOnnxConverter

Constructor

AccelOnnxConverter()

Methods

convert_tf_model

Convert a TensorFlow model into a AccelOnnx model.

convert_tf_model

Convert a TensorFlow model into a AccelOnnx model.

static convert_tf_model(workspace, source_model, input_node, outputs_nodes, require_fpga_conversion=True)

Parameters

workspace

Workspace

Required

Workspace object containing the registered model to retrieve

source_model

model or <xref:<xref:model id str>>

Required

Registered model to be converted

input_node

str

Required

Name of the input layer

outputs_nodes

<xref:<xref:str list>>

Required

List of graph's output nodes where each one will represent the output layer of the network

require_fpga_conversion

bool

default value: True

Specify whether a FPGA is required for processing. Default value is True.

Returns

Object to wait for conversion and get the converted model

Return type

<xref:ModelConvertOperation>