Group17WPIMLDO24/Case-Study-1

1# external imports2from transformers import pipeline3from huggingface_hub import InferenceClient4 5# local imports6import config7 8class Blip_Image_Caption_Large:9    def __init__(self):10        pass11 12    def caption_image(self, image_path, use_local_caption):13        if use_local_caption:14            return self.caption_image_local_pipeline(image_path)15        else:16            return self.caption_image_api(image_path)17    18    def caption_image_local_pipeline(self, image_path):19        self.local_pipeline = pipeline("image-to-text", model=config.IMAGE_CAPTION_MODEL)20        result = self.local_pipeline(image_path)[0]['generated_text']21        return result22 23    def caption_image_api(self, image_path):24        client = InferenceClient(config.IMAGE_CAPTION_MODEL, token=config.HF_API_TOKEN)25        try:26            result = client.image_to_text(image_path).generated_text27        except Exception as e:28            result = f"Error: {e}"29        return result