Model Exchange To ONNX Format

Open Neural Network Exchange (ONNX) is an open ecosystem that empowers AI developers to choose the right tools as their project evolves. ONNX provides an open source format for AI models, both deep learning and traditional ML. It defines an extensible computation graph model, as well as definitions of built-in operators and standard data types. Enabling interoperability between different frameworks and streamlining the path from research to production helps increase the speed of innovation in the AI community.

This program is a secondary development project based on onnx and the work of my research group.

Support Model

Framework / Tool	Model Format	Argparse	Notes
TensorFlow	.pb	1-4, 6, 8
TensorFlow	.ckpt.meta	1-4, 6, 8
TensorFlow	saved_model	1-4, 6-8
Keras	.h5	1-2, 6, 8
MXNet (Apache)	.params & .json	1-2, 5-6, 8
Caffe	.prototxt & cafffemodel	1-2, 6, 8	for Linux/macos
PyTorch	.pth	1-2, 5-6, 8	require NN class

Argparse:

Index	Name	Type	Default	Note
1	model	STRING	None	init FP32 saved model
2	output	STRING	"./output/model.onnx"	output model file
3	input_name	STRING_LIST	["input:0"]	model input_names
4	output_name	STRING_LIST	["result:0"]	model input_names
5	input_size	LIST	[1,3,224,224]	input data size
6	validate	BOOL	True	validate onnx model
7	signature_def	STRING	None	signature_def from saved model
8	opset	INT	7	opset version to use for onnx

Requirements

We need to install some libraries required

pip install -r requirements.txt

Generate model for test

• TensorFlow

  python ./generate/generate_tfmodel.py --mtype=ckpt   # ckpt or pb or saved_model
  

• Keras

  python ./generate/generate_keras.py
  

• Pytorch

  python ./generate/generate_pth.py
  

• MXNet

  python ./generate/generate_mxnet.py
  

Convert

• TensorFlow

  .pb and ckpt

  python convert2onnx.py --model="./model/ckpt/model.ckpt.meta" --output="./output/model.onnx" --input_name=["input:0"] --output_name=["result:0"] --validate=True --opset=7
  

  saved_model

  python convert2onnx.py --model="./model/saved_model" --output="./output/model.onnx" --input_name=["input:0"] --output_name=["result:0"] --signature_def=None --validate=True --opset=7
  

• Keras

  python convert2onnx.py --model="./model/mnistmodel.h5" --output="./output/model.onnx" --validate=True --opset=7
  

• MXNet

  python convert2onnx.py --model="./model/mxnet-model" --output="./output/model.onnx" --input_size=[1,3,3,224] --validate=True --opset=7
  

• Caffe

  python convert2onnx.py --model="./model/caffe-model" --output="./output/model.onnx" --validate=True --opset=7
  

• Pytorch

  python convert2onnx.py --model="./model/model.pth" --output="./output/model.onnx" --input_size=[1,3,3,224] --validate=True --opset=7
  

Reference

• ONNX Tutorials