How to organize data with two classes

vision
1

Hi,

I am currently constructing a semi adversarial auto-encoder. Thus I have two trained networks, one classifier (Male and Female) and one Authenticator (which subject -> Around 450 classes). Because I have to find the two classes of one item I need to strategically organize my data. My question is how can I label the data the best such that is not that much of a hassle?

Is it a good idea to create my own pytorch data base with the following structure:
{‘image’ : image , [‘gender’ : string, ‘personID’:string] }

Thanks in advance
Kind regards

2

Your approach is of course valid and should work when you are properly parsing the strings to classes.
Alternatively, you could also store the class indices directly in your structure.

Also, take a look at e.g. pandas which provide an easy way to store structured data.