I have a model which has been trained with four classes extracted from a secuence in a film. The classes are Frodo, Gandalf, Others and Noone.

When I re-train it with a new Dataset, coming from another secuence of the same film, with this classes, Gandalf, Saruman and Noone, it deletes the classes Frodo and Others, leaving just Gandalf, Saruman and Noone.

So, when I try to test the first secuence once again, it says there is an error, as shown in the image.

What can I do to solve this ?

I’m trying to generate a model that could be used for different secuences of a film, so I would like to know how can I solve this to keep going !