app.py

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

# Step 1: Load Fine-Tuned Model from Hugging Face Model Hub
model_name = "quadranttechnologies/retail-content-safety-clip-finetuned"

print("Loading the fine-tuned model from Hugging Face Model Hub...")
try:
    model = CLIPModel.from_pretrained(model_name, trust_remote_code=True)
    processor = CLIPProcessor.from_pretrained(model_name)
    print("Model loaded successfully.")
except Exception as e:
    print(f"Error loading model or processor: {str(e)}")
    raise

# Step 2: Define the Inference Function
def classify_image(image):
    """
    Classify an image as 'safe' or 'unsafe' with the corresponding percentage.

    Args:
        image (PIL.Image.Image): The input image.
    
    Returns:
        dict: A dictionary containing probabilities for 'safe' and 'unsafe' or an error message.
    """
    try:
        # Check if the image is valid
        if image is None:
            raise ValueError("No image provided. Please upload an image.")
        if not hasattr(image, "convert"):
            raise ValueError("Uploaded file is not a valid image. Please upload a valid image (JPEG, PNG).")

        # Define main categories
        main_categories = ["safe", "unsafe"]

        # Process the image
        print("Processing the image...")
        inputs = processor(text=main_categories, images=image, return_tensors="pt", padding=True)
        print("Inputs processed successfully.")

        # Perform inference
        outputs = model(**inputs)
        print("Model inference completed.")

        # Calculate probabilities
        logits_per_image = outputs.logits_per_image  # Image-text similarity scores
        probs = logits_per_image.softmax(dim=1)  # Convert logits to probabilities

        # Extract probabilities for "safe" and "unsafe"
        safe_probability = probs[0][0].item() * 100  # Convert to percentage
        unsafe_probability = probs[0][1].item() * 100  # Convert to percentage

        print(f"Safe: {safe_probability:.2f}%, Unsafe: {unsafe_probability:.2f}%")

        # Return results
        return {
            "safe": f"{safe_probability:.2f}%",
            "unsafe": f"{unsafe_probability:.2f}%"
        }

    except Exception as e:
        print(f"Error during inference: {str(e)}")
        return {"Error": str(e)}

# Step 3: Set Up Gradio Interface
iface = gr.Interface(
    fn=classify_image,
    inputs=gr.Image(type="pil"),
    outputs=gr.Label(label="Output"),  # Use Gradio's Label component for user-friendly display
    title="Content Safety Classification",
    description="Upload an image to classify it as 'safe' or 'unsafe' with corresponding probabilities.",
)

# Step 4: Launch Gradio Interface
if __name__ == "__main__":
    iface.launch()