As a part of an industrial IoT solution, you have four high-speed cameras used for quality control, and connected to an edge device. The images sent by the cameras must be pre-processed locally on the edge, and only images with identified quality defects must be forwarded to the IoT hub for further analysis and archiving. You need to create three IoT Edge modules that receive messages from the cameras, run a machine learning model to process images, then forward the relevant images to IoT Hub. You need to deploy two custom modules on your device. 
Which property groups should you add to the deploymentManifest.json?

Answer options:

A.“$edgeAgent.systemModules” and “$edgeAgent.properties.desired.routes”
B.“$edgeAgent.modules” and “$edgeHub.properties.desired.routes”
C.“$edgeHub.modules” and “$edgeHub.properties.desired.routes”
D.“HostConfig.Binds” and "registryCredentials"

Correct Answer: B
Option A is incorrect because “$edgeAgent.systemModules”is used to define properties of the standard edge modules like edgeAgent and edgeHub. Definitions of custom modules go under the “modules” property.
Option B is CORRECT because the custom module definitions must go into the “modules” section of the edgeAgent, while the connections between the modules are described in the “routes” section of the edgeHub definition.
Option C is incorrect because the custom module definitions must go into the “modules” section of the edgeAgent, rather than the edgeHub.
Option D is incorrect because the “Binds” parameter is needed if a module needs to access a local folder (typically for storing data locally). It is not needed for the current scenario. registryCredentials is actually needed and used to define the custom container registries.
References: 

https://docs.microsoft.com/en-us/azure/iot-edge/tutorial-machine-learning-edge-06-custom-modules?view=iotedge-2018-06
https://github.com/Azure-Samples/IoTEdgeAndMlSample