The prediction response we get from the cloud server will vary with the Tensorflow code we use for the project. To give an insight,
A classification problem will respond as
{ “Class1” : percentage, “Class2” : precentage, …}
Image caption will respond as
{ "caption": [ "the", "quick", "brown", "", "", "" ] }
I have asked a StackOverFlow question on this.

There are a variety of ways we can get the prediction response. We can add a few prediction formats with the module and give the user flexibility to create custom plugins to parse the prediction output.

Comments

boaloysius created an issue.