Abstract: We study a novel problem in this manuscript, that is, if a modern text-to-image diffusion model can tailor any image classifier across domains and categories. Existing domain adaption works ...
Some results have been hidden because they may be inaccessible to you
Show inaccessible results