cailean
/
ofxOnnxRuntime
1
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity
Onnxruntime (1.21) for openFrameworks (0.12) using CUDA (12.8) & CUDNN (9.7)
You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
14 Commits
2 Branches
0 Tags
532 MiB
 
 
Tree: 86d7377ad1
Branches Tags
${ item.name }
Create tag ${ searchTerm }
Create branch ${ searchTerm }
from '86d7377ad1'
${ noResults }
ofxOnnxRuntime/README.md

1.6 KiB
Raw Blame History

ofxOnnxRuntime

Updated version, working with Windows 11, CUDA 12.8, and ONNXRuntime 1.20.1

It is not the cleanest implementation, but it works!

In this implementation I have added the ability to batch process. This will only apply to models that allow it.

ONNX Runtime tiny wrapper for openFrameworks

'test'

Installation

  • macOS
    • copy libonnxruntime.1.10.0.dylib to /usr/local/lib
    • Generate a project using ProjectGenerator.
  • Windows
    • There are two ways to install ONNX Runtime on your project.
    1. Install using NuGet
      • I recommend this way in general.
      • Generate a project using ProjectGenerator.
      • Open sln file.
      • Right click your project on Solution Explorer pane, and then select Manage NuGet Packages....
      • From Browse tab, search Microsoft.ML.OnnxRuntime (CPU) or Microsoft.ML.OnnxRuntime.Gpu (GPU) and install it.
    2. DLL direct download
      • You can download prebuilt DLLs from here.
      • Unzip downloaded onnxruntime-win-x64-(gpu-)1.20.1.zip and locate files on libs\onnxruntime\lib\vs\x64\ .
      • Generate a project using ProjectGenerator, then all libs are linked correctly and all dlls are copied to bin.

Tested environment

  • oF 0.11.2 + MacBookPro 2018 Intel + macOS Catalina
  • oF 0.11.2 + VS2017 + Windows 10 + RTX2080Ti + CUDA 11.4

ToDo

  • check M1 Mac (should work), Linux CPU&GPU

Reference Implementation