simple-onnx-processing-tools

A set of simple tools for splitting, merging, OP deletion, size compression, rewriting attributes and constants, OP generation, change opset, change to the specified input order, addition of OP, RGB to BGR conversion, change batch size, batch rename of OP, and JSON convertion for ONNX models.

1. Tools

HostPC

# (1) Minimum configuration installation with no dependent packages installed
$ pip install -U simple-onnx-processing-tools \
&& pip install -U onnx \
&& python3 -m pip install -U onnx_graphsurgeon --index-url https://pypi.ngc.nvidia.com

or

# (2) When installing all dependent packages such as onnx-simplifier, onnxruntime, numpy, etc...
$ pip install -U simple-onnx-processing-tools[full] \
&& pip install -U onnx \
&& python3 -m pip install -U onnx_graphsurgeon --index-url https://pypi.ngc.nvidia.com

Docker

$ docker run --rm -it \
-v `pwd`:/workdir \
-w /workdir \
pinto0309/simple-onnx-processing-tools:1.0.42

No.	Tool Name	Tags	Summary
1	snc4onnx		Simple tool to combine(merge) onnx models. Simple Network Combine Tool for ONNX.
2	sne4onnx		A very simple tool for situations where optimization with onnx-simplifier would exceed the Protocol Buffers upper file size limit of 2GB, or simply to separate onnx files to any size you want. Simple Network Extraction for ONNX.
3	snd4onnx		Simple node deletion tool for onnx. Simple Node Deletion for ONNX.
4	scs4onnx		A very simple tool that compresses the overall size of the ONNX model by aggregating duplicate constant values as much as possible. Simple Constant value Shrink for ONNX.
5	sog4onnx		Simple ONNX operation generator. Simple Operation Generator for ONNX.
6	sam4onnx		A very simple tool to rewrite parameters such as attributes and constants for OPs in ONNX models. Simple Attribute and Constant Modifier for ONNX.
7	soc4onnx		A very simple tool that forces a change in the opset of an ONNX graph. Simple Opset Changer for ONNX.
8	scc4onnx		Very simple NCHW and NHWC conversion tool for ONNX. Change to the specified input order for each and every input OP. Also, change the channel order of RGB and BGR. Simple Channel Converter for ONNX.
9	sna4onnx		Simple node addition tool for onnx. Simple Node Addition for ONNX.
10	sbi4onnx		A very simple script that only initializes the batch size of ONNX. Simple Batchsize Initialization for ONNX.
11	sor4onnx		Simple OP Renamer for ONNX.
12	soa4onnx		Simple model Output OP Additional tools for ONNX.
13	ssi4onnx		Simple Shape Inference tool for ONNX.
14	sit4onnx		Tools for simple inference testing using TensorRT, CUDA and OpenVINO CPU/GPU and CPU providers. Simple Inference Test for ONNX.
15	onnx2json		Exports the ONNX file to a JSON file.
16	json2onnx		Converts a JSON file to an ONNX file.
17	sed4onnx		Simple ONNX constant encoder/decoder. Since the constant values in the JSON files generated by onnx2json are Base64-encoded values, ASCII <-> Base64 conversion is required when rewriting JSON constant values.
18	ssc4onnx		Checker with simple ONNX model structure. Simple Structure Checker for ONNX. Analyzes and displays the structure of huge size models that cannot be displayed by Netron.
19	sio4onnx		Simple tool to change the INPUT and OUTPUT shape of ONNX.
20	components_of_onnx	[WIP]	ONNX parts yard. The various operations described in Operator Schemas are converted in advance into OP stand-alone ONNX files.

2. Very useful tools

No.	Tool Name	Author	Tags	Summary
1	OnnxGraphQt	fateshelled		ONNX model visualizer. Model structure can be edited on the visualization tool.
2	onnx-simplifier	daquexian		ONNX Simplifier is presented to simplify the ONNX model. It infers the whole computation graph and then replaces the redundant operators with their constant outputs.
3	Sparsify	neuralmagic		Easy-to-use UI for automatically sparsifying neural networks and creating sparsification recipes for better inference performance and a smaller footprint.
4	DeepSparse Engine	neuralmagic		Sparsity-aware neural network inference engine for GPU-class performance on CPUs.
5	Sparsebit	megvii-research		Sparsebit is a toolkit with pruning and quantization capabilities. It is designed to help researchers compress and accelerate neural network models by modifying only a few codes in existing pytorch project.

2-1. OnnxGraphQt - [WIP] Startup Method Sample

$ xhost +local: && \
docker run -it --rm \
-v `pwd`:/home/user/workdir \
-v /tmp/.X11-unix/:/tmp/.X11-unix:rw \
--net=host \
-e XDG_RUNTIME_DIR=$XDG_RUNTIME_DIR \
-e DISPLAY=$DISPLAY \
--privileged \
ghcr.io/pinto0309/openvino2tensorflow:latest

$ git clone https://github.com/fateshelled/OnnxGraphQt \
&& cd OnnxGraphQt \
&& sudo python3 -m pip install -r requirements.txt -U \
&& cd ..

$ python3 OnnxGraphQt/onnxgraphqt/main.py

3. Acknowledgments

1. https://github.com/onnx/onnx/blob/main/docs/PythonAPIOverview.md
2. https://docs.nvidia.com/deeplearning/tensorrt/onnx-graphsurgeon/docs/index.html
3. https://github.com/NVIDIA/TensorRT/tree/main/tools/onnx-graphsurgeon
4. https://github.com/onnx/onnx/blob/main/docs/Operators.md

4. References

1. https://github.com/PINTO0309/PINTO_model_zoo