MIVisionX toolkit is a set of comprehensive computer vision and machine intelligence libraries, utilities, and applications bundled into a single toolkit. AMD MIVisionX also delivers a highly optimized open-source implementation of the Khronos OpenVX™ and OpenVX™ Extensions.

Image Clasiification using AMD WinML Extension

This application implements a method to run WinML supported ONNX models using MIVisionX RunTime. This c/c++ implementation does image classification on live camera capture or pre-recorded video stream.

This application is a sample for developing windows application using MIVisionX WinML extension.

Build Instructions


Build using Visual Studio 2019 on 64-bit Windows 10 / 11


Step 1. Get ONNX model

Step 2. Build the app using winml_classifier.sln on Visual Studio.

Step 3. Run tests


 .\winml_classifier.exe --inception  <FULL_PATH_TO\inception_v2\model.onnx> [OPTIONAL]  
                        --resnet50   <FULL_PATH_TO\resnet50\model.onnx>     [OPTIONAL] 
                        --vgg19      <FULL_PATH_TO\vgg19\model.onnx>        [OPTIONAL] 
                        --shufflenet <FULL_PATH_TO\shufflenet\model.onnx>   [OPTIONAL] 
                        --squeezenet <FULL_PATH_TO\squeezenet\model.onnx>   [OPTIONAL] 
                        --densenet   <FULL_PATH_TO\densenet\model.onnx>     [OPTIONAL] 
                        --zfnet      <FULL_PATH_TO\zfnet\model.onnx>        [OPTIONAL] 
                        --label FULL_PATH_TO\amd_openvx_extensions\amd_winml\samples\data\Labels.txt [REQUIRED]
                        --capture 0/ --video <video file>  [REQUIRED]

MIVisionX Image Classification

MIVisionX Image Classification

MIVisionX Image Classification using WinML

MIVisionX Image Classification using WinML


 .\winml_classifier.exe --inception FULL_PATH_TO\inception_v2\model.onnx
                        --resnet50  FULL_PATH_TO\resnet50\model.onnx 
                        --zfnet FULL_PATH_TO\zfnet\model.onnx 
                        --label FULL_PATH_TO\MIVisionX\amd_openvx_extensions\amd_winml\samples\data\Labels.txt
                        --capture 0