blip_caption.py

from transformers import BlipProcessor, BlipForConditionalGeneration
from PIL import Image

# Load once on startup for performance
processor = BlipProcessor.from_pretrained("Salesforce/blip-image-captioning-base")
model = BlipForConditionalGeneration.from_pretrained("Salesforce/blip-image-captioning-base")

def generate_caption(pil_img):
    inputs = processor(images=pil_img, return_tensors="pt")
    out = model.generate(**inputs)
    caption = processor.decode(out[0], skip_special_tokens=True)
    return caption.lower()