Method visual_prompt_inference

Visual prompt inference for both interactive and generic workflow. Args: target_image (Union[str, Image.Image]): The image to upload. Can be a file path or PIL.Image prompts (List[dict]): A list of prompt dict. Each dict is for one prompt image: # Bo

(
        self,
        target_image: Union[str, Image.Image],
        prompt: List[Dict],
        return_type: List[str] = ["bbox"],
    )