Image classification with new class

Hi there,
I trained a ResNet34 model to classify the given images into three classes. Now I want to add a new class into my model. Do I need to train my whole model again on all four classes or is there any way I can just train my model on new class and get the predictions for all four classes?


Hi My!

If you literally mean train your model with batches that consist
solely of samples of the new class, then, no, this won’t work. As
it trains, your model will simply learn to predict your new class
all of the time, regardless of the input, and, with respect to your
single-class training set, it will always be right.

If you are asking something else, please clarify.


K. Frank

Thanks for the reply. Yes, I mean training the model with batches of samples of new class.
I am wondering if there is any way to train the existing model (already trained on three classes) further on new class samples, and finally getting the model trained for all four classes.