Azure Cognitive Services are great, but you need a reliable network connection to the cloud, right? Wrong! For one thing, there's an increasing number of services that can be deployed to containers. If you don't have a reliable network connection, or have limited bandwidth, or if you have data that's needs to be kept on-premises, these are all use-cases for deploying Azure Cognitive Services to containers. We can also use Azure IoT Hub and IoT Edge to manage devices and services. A second option is to export a custom trained model. We'll see how to use Custom Vision in Azure Cognitive Services to train a custom computer vision model and export it as a TensorFlow model, and then deploy it to a Raspberry Pi Zero single board computer. Either way, you get the advantages of Azure Cognitive Services, but running outside Azure!
You will learn:
- How to deploy Azure Cognitive Services as containers
- How to use Azure IoT Hub to manage devices
- How to export a Custom Vision model to a device