The simple answer is yes. But the method may be complex depending on the format of your data - what program your labels were made with. Essentially deeplabcut just wants x, y (pixel) coordinates.
The easiest way to see the format is to run the test script, and open the labeled data folder to look at one of the csv’s. Or start a project, name the body parts you would track, label a few images, and look at the labeled data. That would provide you with a template for the format.
If you can get the data into the right format, you can (in deeplabcut) add a new video to your project, extract frames, and then open and edit the blank csv to add your coordinates. Then you can run
deeplabcut.convertcsv2h5(...) to convert the csv file you made to the h5 needed to make the training set.
Some users (including me) have written scripts to import labels from different formats, so feel free to post your labelling source and you might find that the hard work has already been done.