I'd like to be able to use hot keys to label images more efficiently. I could use Image Captioning, but in order to save the data, it seems that a mouse click is necessary.
Hi! By image classification, do you mean an interface like this one where you assign one or more labels to images? https://prodi.gy/docs/computer-vision#classification-multi
If so, you should definitely be able to do the whole end-to-end annotation process via hotkeys 
By default, the choice options are mapped to the number keys, e.g. 1 for the first option. But you can also customize the shortcuts, e.g. {"1": "X", "2": "Y"} to map the first option to X and so on.
If your classification labels are mutually exclusive, you can also set "choice_auto_accept": true in the "config" returned by your recipe to auto-save once an option was selected. Otherwise, you can press "accept" (A or a custom shortcut) to accept and save the annotation.