NA-MIC Project Weeks

Back to Projects List

Using ONNX runtime to facilitate the usage of PyTorch3D models in Windows

Key Investigators

• Sal Choueib (Ebatinca, Spain)
• Csaba Pinter (Ebatinca, Spain)
• Juan Ruiz Alzola (Ebatinca, Spain)

Presenter location: Online

Project Description

The ONNX runtime is an intermediary machine learning framework that allows users to easily convert between different machine learning frameworks. The aim in this project is to leverage ONNX to utilize PyTorch3D models in Windows.

Objective

1. Build ONNX on a windows machine
2. Use the ONNX runtime to load and run the following PyTorch model in windows: https://github.com/DCBIA-OrthoLab/SlicerDentalModelSeg
3. Use the runtime to tune the performance of the model in windows
4. Deploy the model

Approach and Plan

1. Investigate the ONNX runtime environment
2. Attempt to export the given PyTorch model in ONNX format
3. Attempt to import the model and run it in the ONNX runtime in windows
4. Investigate the performance tuning capabilities of ONNX
5. Outline a pipeline to streamline the conversion of PyTorch models to be used on windows systems

Progress and Next Steps

No response

Illustrations

No response

Background and References

No response