Put glasses on a dog. Sometimes this breaks when the Google image API changes. TODO: specify an API version if possible
Dog score only supports .jpg, I think because the ML model was trained on jpeg data.
TODO: Keyboard support / a11y
Should work on latest builds of Chrome and Firefox.