Hi!
We are developing app using Core ML. We have model written on Keras and exported with coreml tools(python library) into .mlmodel file. Review of this model says that model exported as expected. Our model is a Convolutional Neural Network that work as classifier
Model paramteres(expected and how it works on Python, and how it works with model exported to Android) :
Input - list of 1000 floats(time serie){1,1000,1}
Output - list of 2 floats([1] - possibility of belonging to the 1-st class, [2] - possibility of belonging to the 2-d class)
Actual result on iOS :
Input - list of 1000 floats(time serie){1,1000,1}
Output - MultiArray< Float 1000,2> - Thousand pairs of 2 floats
Interesting that model on iOS always classify data to 2-d class(every pair in list) even when 1-st class expected. And it looks like this output always the same even for different data.
