Classes are pre-defined types of your annotations (i.e. "Background" or "Tumor"). Thus, every labeled object on image or video frame has exactly one class associated.
Usually, you define classes before labeling starts, but you can add new classes in the labelling interface on-fly (if you have enough permissions, of course).
You can limit tools that can be used to create an instance of a class by setting a class Shape. For example, if you select shape "Bitmap", labeler can an object using Brush or Smart Tool, but not Polygon. This can be very helpful in big projects and guarantee thet your data scientists won't need to convert Cuboids to Polygons two months after labeling is finished.
If you don't to limit shape, set it to Any Shape — thus you can have "Cars" made of both Polygons and Bitmaps.
Optionally, you can assign a Hotkey for a class to quickly select it during labeling. You can only set a single latin character (because other combinations may be unavailable).