Replace the .pt trained network

[Delphi89]

Hi,

I hope this is not too much off topic.

I have a different trained PyTorch network that recognises only pipes. I have replaced the mobilenet_quantized.pt file with my .pt network file and words.txt with a file containing only one object: pipe.

However, I'm not sure what your score means, with my network is negative most of the time. Could you please explain how I can get the pure probability of object recognition and object coordinates from the inference?

Thank you! Paul