How to deploy my Custom Vision model for Object Detection

Question

Hi I was wondering how I could deploy my Custom Vision Model which I have just finished training. I would like to know if it is easier to deploy using Colaboratory or to use Azure. Also if Azure is the simpler and more convenient way to do so, could you guide me on the method please.

Also, is it possible to integrate live video object detection for my model?

Answer 1

To deploy your Custom Vision model for Object Detection, using Azure is generally considered a more straightforward and integrated approach compared to using Colaboratory. Azure provides a dedicated service for deploying machine learning models, which simplifies the process.

Here’s a brief guide on how to deploy your Custom Vision model using Azure:

1. Register Your Model: First, ensure that your trained model is registered in the Azure Custom Vision service.
2. Create a Deployment: You can deploy your model directly from the Custom Vision portal. After selecting your model, choose the option to deploy it as an endpoint.
3. Select Deployment Options: You can choose to deploy to Azure Kubernetes Service or Azure Container Instance. You will need to configure the endpoint with necessary parameters such as the entry script and dependencies.
4. Deploy the Model: Once you have configured the settings, you can initiate the deployment. After successful deployment, you will receive a REST endpoint that you can use for predictions.

Regarding live video object detection, it is possible to integrate this functionality with your deployed model. You would typically set up a system that captures video frames, sends them to the deployed model for inference, and processes the results to detect objects in real-time.

---

References:

• What is Custom Vision?
• Quickstart: Custom text classification (language-studio)
• Use the studio to deploy models trained in the designer