How to get the output of a trained model (deep learning), object recognition and image segmentation

Hi all,

I would like to get the output of a trained model (Deep Learning) for object recognition and image segmentation tasks. I am looking for the right way how to use such a model in Unity that I have already trained with PyTorch or Tensorflow.
Could anyone point me to the right direction? This would be great.
What is the general workflow? What kind of libraries should I use?

Is the ML Agents Toolkit a solution? As far as I am understand , it is "only" Deep Reinforcement Learning which solves Markov Decision Process problems and which is rather not related to object recognition and image segmentation tasks. I might be wrong, however.

 

did you find an answer for that?

 

@sa2706 I have not pursued, but want to try again with Unity Barracuda. With Barracuda it is now possible to import Pytorch, Tensorflow models that have been converted to ONNX format.