marrying grounding dino and stable diffusion

Object detection has traditionally been a closed-set problem: you train on a fixed list of classes and cannot recognize new ones. Grounding DINO breaks this mold, becoming an open-set, language-conditioned