Hi guys,
I 'm new in caffe and i use it for my thesis. I've some questions:
1)I'd like to know what exactly I put in "label". For my NN I use HDF5 Data Layer.
In
top: "data" = I put the training data;
but in:
top: "label": what do I must put? I don't understand if i put the label of the class of which my training data belongs (is the samething of the groundthruth)
2) In the Innerproduct layer does the number of outputs match the number of my classes? (For example i have two classes like "dog" and "cat" and my number of outputs is 2?
3) I've read that to set the number of classes i must write two file "synsets.txt" to declare the number of classes and "synset_word.txt" to give othe description (for example "0 dog" and "1 cat"). Is it correct? And How my NN takes in in imput these files?
Please someone help me!!
thanks a lot,
Gianni