3D daya to cobtibious (1,3) network?

1

I have 3D grayscale photo data of 6 slices and my input shape is therefore (6,1,100,100)

I want to estimate realtime 3 different characteristics and therefore (1,3) shaped output. The estimated values are all continious variables.

which network would be suitable dor estimation of continious variables?

2

A CNN would probably be a good option. Just don’t use an activation function at the end so the output is continuous. You could squeeze the image to get rid of the 1 dimension and use 6 as your input channels. This would give you the correct output.