Insiders Hub

PyTorch2OpenVINO

Aven Le Zhou

Aven Le Zhou

Shanghai, Shanghai

0 1
  • 0 Collaborators

All credits go to INTEL AI Academy support team. Special thanks go to Ellick Chan and Huiyan Cao. This is a short tutorial shows how to port pre-trained PyTorch model to INTEL OpenVINO model. In short, the pre-trained PyTorch model got converted to ONNX format and then optimised by OpenVINO model optimiser. ...learn more

Project status: Published/In Market

Artificial Intelligence, Graphics and Media

Intel Technologies
AI DevCloud / Xeon, Intel Python, MKL, OpenVINO

Code Samples [1]

Overview / Usage

We are trying to port PyTorch Pre-trained Model to INTEL OpenVINO for Inference and we used ShanshuiDaDA Model as Example.

Methodology / Approach

a. Convert pre-trained pytorch model (latest_net_G.pth) to ONNX (lastest_net_G.onnx) with pytorch2onnx.py

b. Convert ONNX model (lastest_net_G.onnx) to IR (onnx_fp32 or onnx_fp16) using mo.sh (with different data type setting in mo.sh)

Technologies Used

OpenVINO

Repository

https://github.com/aaaven/PyTorch2OpenVINO

Comments (1)