import gradio as gr
from transformers import pipeline
from PIL import Image

# Define the image classification function
def classify_image(image):
    try:
        # Convert the Gradio image input (which is a NumPy array) to a PIL image
        image = Image.fromarray(image)

        # Create the image classification pipeline
        img_class = pipeline(
            "image-classification", model="AMfeta99/vit-base-oxford-brain-tumor_x-ray"
        )

        # Perform image classification
        results = img_class(image)

        # Find the result with the highest score
        max_score_result = max(results, key=lambda x: x['score'])

        # Extract the predicted label
        predictions = max_score_result['label']
        
        if predictions==1:
          text_pred='Tumor'
        else:
          text_pred='Normal'

        return text_pred

    except Exception as e:
        # Handle any errors that occur during classification
        return f"Error: {str(e)}"

# Define the Gradio interface
image = gr.Image()
label = gr.Label(num_top_classes=1)
title = "Brain Tumor Classification"
description = "Worried about whether your brain scan is normal or not? Upload your brain scan and the algorithm will give you an expert opinion. Check out [the original algorithm](https://huggingface.co/AMfeta99/vit-base-oxford-brain-tumor) that this demo is based of."
article = "<p style='text-align: center'>Image Classification | Demo Model</p>"


# Prepare examples with loaded images
examples = ["Normal_0.jpg", "Normal_1.jpg", "Normal_2.jpg", "Tumor_0.jpg", "Tumor_1.jpg", "Tumor_2.jpg"]

demo = gr.Interface(fn=classify_image, inputs=image, outputs=label, description=description, article=article, title=title, examples=examples)

# Launch the Gradio interface
demo.launch(share=True)