Make predictions with ONNX on computer vision models from AutoML on Azure IoT Edge (ARM32 device)
- Connect to an edge device.
- Clone this repository.
git clone https://github.com/yahanda/automl-image-prediction-with-onnx-on-edge.git
- Download
model.onnx
andlabels.json
from an AutoML training run, and put them under the folder<cloned-repository>/automl-image-prediction-with-onnx-on-edge/app/
. - Build and push docker image.
cd automl-image-prediction-with-onnx-on-edge/dockerfiles sudo git submodule update --init sudo docker build -t automlimageprediction -f Dockerfile.arm32v7 .. sudo docker tag automlimageprediction <your-container-registry>/automlimageprediction:<version>-arm32v7 sudo docker push <your-container-registry>/automlimageprediction:<version>-arm32v7
- Deploy the module to Azure IoT Edge device