Classifier Image

Hello there..
I am currently working on a project to develop an image classification application. I have created six variables with the theme of shirts, and I have already used classify image with a loss of 0.02 on the training data. However, I am confused about what to do when testing the camera application. If the camera is not pointing at a shirt but at something random, how can I show a notification that the camera does not detect a shirt? Should I add datasets of non-shirt images? What are the steps to do that, or is there another way?

See the example linked to, which looks at different weather conditions on a mountain

Does this help ?

It helps, so should I add a dataset with the variable 'that is not a shirt'?
Are there any other ways?