accuracy of all labels

[antonio cayulao]

Hi,

If I have a 'lenet_iter_10000' trained, How can I do to extract the accuracy of all labels that  with I trained this?

For example I see the final of this:

https://github.com/gnperdue/ANNMINERvA/blob/master/understanding_uv_only_for11seg.ipynb

But all is done from 'perfmat1460659054.npy' If I don't have this file, Is there some way to make the extract all accuracy of labels?

Greetings!

[Ahmed Ibrahim]

I did not understand the question well , but if you are talking about per class accuracy you can do something like this 

layer {
  name: "accuracy"
  type: "Accuracy"
  bottom: "fc8"
  bottom: "label"
  top: "accuracy"
  top: "class"  # this will print accuracy per class
  include {
    phase: TEST
  }
}