app.py

import gradio as gr
from transformers import AutoProcessor, AutoModelForImageTextToText
from PIL import Image
import torch

# Load once
processor = AutoProcessor.from_pretrained("RSRathore/ny-image-captioning")
model = AutoModelForImageTextToText.from_pretrained("RSRathore/ny-image-captioning")
model.eval()

def generate_caption(image):
    inputs = processor(images=image, return_tensors="pt").to(model.device)
    with torch.no_grad():
        outputs = model.generate(**inputs)
    caption = processor.batch_decode(outputs, skip_special_tokens=True)[0]
    return caption

demo = gr.Interface(
    fn=generate_caption,
    inputs=gr.Image(type="pil"),
    outputs="text",
    title="🖼️ Comical Caption Generator",
    description="Upload an image (png or jpg) and get a caption using the `RSRathore/ny-image-captioning` model."
)

demo.launch()