Abstract: This paper proposes a unified multi-modal image generation method to address two critical challenges in visible-infrared (VI) person re-identification (ReID): the insufficiency of training ...