app.py

import gradio as gr
from transformers import CLIPProcessor, CLIPModel
from PIL import Image
import torch

model = CLIPModel.from_pretrained("openai/clip-vit-base-patch32")
processor = CLIPProcessor.from_pretrained("openai/clip-vit-base-patch32")

def generate_caption(image):
    if image is None:
        return "No image uploaded."

    # Candidate text prompts
    texts = [
        "a photo of a cat",
        "a photo of a dog",
        "a photo of a man",
        "a photo of a woman",
        "a photo of a laptop",
        "a photo of a smartphone",
        "a photo of a city",
        "a photo of a landscape",
        "a photo of food",
        "a photo of a car"
    ]

    inputs = processor(text=texts, images=image, return_tensors="pt", padding=True)
    outputs = model(**inputs)

    logits_per_image = outputs.logits_per_image  # image-text similarity scores
    probs = logits_per_image.softmax(dim=1)      # convert to probabilities

    best_match = torch.argmax(probs).item()
    caption = texts[best_match]
    return f"Best match: {caption} (Confidence: {probs[0][best_match].item():.2f})"

iface = gr.Interface(
    fn=generate_caption,
    inputs=gr.Image(type="pil"),
    outputs=gr.Textbox(label="Generated Caption"),
    title="Image Captioning with CLIP",
    description="Upload an image and get a dynamically generated caption using CLIP.<br><br><b>Detectable categories:</b><br><div style='display: flex; gap: 40px;'><div><ul><li>a photo of a cat</li><li>a photo of a dog</li><li>a photo of a man</li><li>a photo of a woman</li><li>a photo of a laptop</li></ul></div><div><ul><li>a photo of a smartphone</li><li>a photo of a city</li><li>a photo of a landscape</li><li>a photo of food</li><li>a photo of a car</li></ul></div></div>"
)

iface.launch()