I Used a Regression CNN for estimating soil moisture based on 5 input images and a numeric SMAP data and the value of soil moisture will always be less than this SMAP input. So I want a way to integrate this output into the model training so that the model can converge quickly and precisely.
So the input data is 512,512,5 numpy array and a SMAP value in m3/m3. The output is a numeric soil moisture value in terms of m3/m3.
Is there a way to ensure this?
