ONNXExplorer

onnxexp is an easy way to explore your onnx model. It helps you debug and even convert/inference your onnx simply with a single command line.

Now, you have another option to view, debug, summary your model without netron. You can also using this tool converts your model to trt engine.

You can install onnxexplorer simply by:

pip install onnxexplorer

then you can using onnxexp -h to see what it capable of:

usage: onnxexp [-h] [--version] {glance,totrt,check} ...

positional arguments:
  {glance,totrt,check}
    glance              Take a glance at your onnx model.
    totrt               Convert your model to trt using onnx-tensorrt python
                        API.
    check               Check your onnx model is valid or not.

optional arguments:
  -h, --help            show this help message and exit
  --version, -v         show version info.

The final function onnxexp will provide are:

• Glance at your onnx model info, print out input and output shapes information, and Node sets;
• Search node detail by provide node id, or node type, etc, search all Conv in your model;
• Int8 convert of your onnx model;
• Convert your model to tensorrt via onnx-tensorrt python API;
• Calculate your model params and test speed via ONNXRuntime;

Update

• 2021.12.22: Add TensorRT convert function, now you can using onnxexp convert your onnx model to trt engine, even with dynamic input models:

  onnxexp totrt -m shufflenetv2_body.onnx --min_shapes img:1x3x384x288 --opt_shapes img:2x3x384x288 --max_shapes img:4x3x384x288
  

• 2021.12.04: Update args, re-organised readme and usage;
• 2021.01.06: Update search functions;
• 2019.09.30: First released this package;

Install

to install onnxexplorer, you can do:

sudo pip3 install onnxexplorer

Or if pip not available:

sudo python3 setup.py install

Copyright

All right reserved by Lucas Jin. Codes released under Apache License.