Closed hanumanraje closed 2 years ago
I haven't tried, but I am quite sure that training the same model on two different classes will generate poor results. A model (i.e. the embedding core set in patchcore or mean and covariance in padim) is meant to be trained on only one class. That is also what they do in the papers that the models are based on.
How to Train it for multiple classes of data like bottle, carpet in same model