ONNX-MXNet API¶
Overview¶
ONNX is an open format to represent deep learning models. With ONNX as an intermediate representation, it is easier to move models between state-of-the-art tools and frameworks for training and inference.
The mxnet.contrib.onnx
package refers to the APIs and interfaces that implement ONNX model format support for Apache MXNet.
With ONNX format support for MXNet, developers can build and train models with a variety of deep learning frameworks, and import these models into MXNet to run them for inference and training using MXNet’s highly optimized engine.
Warning
This package contains experimental APIs and may change in the near future.
Installation Instructions¶
- To use this module developers need to install ONNX, which requires the protobuf compiler to be installed separately. Please follow the instructions to install ONNX and its dependencies. MXNet currently supports ONNX v1.2.1. Once installed, you can go through the tutorials on how to use this module.
This document describes all the ONNX-MXNet APIs.
mxnet.contrib.onnx.onnx2mx.import_model |
Functions for importing ONNX models to MXNet and for checking metadata |
mxnet.contrib.onnx.onnx2mx.import_to_gluon |
Import ONNX model to gluon interface |
mxnet.contrib.onnx.mx2onnx.export_model |
Exports an MXNet model to the ONNX model format |
ONNX Tutorials¶
ONNX Examples¶
- Face Recognition with ArcFace
- Image Classification with MobileNet, ResNet, SqueezeNet, VGG
API Reference¶
Functions for importing ONNX models to MXNet and for checking metadata
-
mxnet.contrib.onnx.onnx2mx.import_model.
import_model
(model_file)[source]¶ Imports the ONNX model file, passed as a parameter, into MXNet symbol and parameters. Operator support and coverage - https://cwiki.apache.org/confluence/display/MXNET/MXNet-ONNX+Integration
Parameters: model_file (str) – ONNX model file name Returns: Notes
This method is available when you
import mxnet.contrib.onnx
-
mxnet.contrib.onnx.onnx2mx.import_model.
get_model_metadata
(model_file)[source]¶ Returns the name and shape information of input and output tensors of the given ONNX model file.
Notes
This method is available when you
import mxnet.contrib.onnx
Parameters: model_file (str) – ONNX model file name Returns: model_metadata – A dictionary object mapping various metadata to its corresponding value. The dictionary will have the following template: 'input_tensor_data' : list of tuples representing the shape of the input paramters 'output_tensor_data' : list of tuples representing the shape of the output of the model
Return type: dict
Import ONNX model to gluon interface
-
mxnet.contrib.onnx.onnx2mx.import_to_gluon.
import_to_gluon
(model_file, ctx)[source]¶ Imports the ONNX model files, passed as a parameter, into Gluon SymbolBlock object.
Parameters: - model_file (str) – ONNX model file name
- ctx (Context or list of Context) – Loads the model into one or many context(s).
Returns: sym_block – A SymbolBlock object representing the given model file.
Return type: Notes
This method is available when you
import mxnet.contrib.onnx
Exports an MXNet model to the ONNX model format
-
mxnet.contrib.onnx.mx2onnx.export_model.
export_model
(sym, params, input_shape, input_type=, onnx_file_path=u'model.onnx', verbose=False)[source]¶ Exports the MXNet model file, passed as a parameter, into ONNX model. Accepts both symbol,parameter objects as well as json and params filepaths as input. Operator support and coverage - https://cwiki.apache.org/confluence/display/MXNET/MXNet-ONNX+Integration
Parameters: - sym (str or symbol object) – Path to the json file or Symbol object
- params (str or symbol object) – Path to the params file or params dictionary. (Including both arg_params and aux_params)
- input_shape (List of tuple) – Input shape of the model e.g [(1,3,224,224)]
- input_type (data type) – Input data type e.g. np.float32
- onnx_file_path (str) – Path where to save the generated onnx file
- verbose (Boolean) – If true will print logs of the model conversion
Returns: onnx_file_path – Onnx file path
Return type: str
Notes
This method is available when you
import mxnet.contrib.onnx