Probablistic Output

[LukeAChaplin]

Hello everyone,

I have a quick, and possibly very naive, question. Firstly, thank you very much for all your work Marco. I am working on a very similar segmentation problem and wondering how I can use predict on the test data to produce masks with values proportional to the certainty of the class? Currently the output is binary.

Any help would be much appreciated,

Luke

[mrgloom]

Output is sigmoid, so it's in [0,1] range, you can interpret it as probability.

[LukeAChaplin]

Thank you very much. I was just being a little dim about how I made my predictions.