import gradio as gr
import numpy as np
from PIL import Image
from tensorflow.keras.models import load_model
from huggingface_hub import hf_hub_download

# Download the model from Hugging Face Hub
model_path = hf_hub_download(repo_id="TaliZG03/kidney_normal_CT_classifier_model", filename="model.keras")

# Load the model
model = load_model(model_path)

# Preprocess function
def preprocess(image):
    image = image.resize((300, 300)).convert("RGB")
    image = np.array(image) / 255.0
    return np.expand_dims(image, axis=0)

# Prediction function with flagging
def predict(image):
    input_array = preprocess(image)
    prediction = model.predict(input_array)[0][0]

    label = "NORMAL" if prediction >= 0.5 else "ABNORMAL"
    confidence = prediction if label == "NORMAL" else 1 - prediction

    if label == "NORMAL" and confidence >= 0.7:
        explanation = " The kidney CT scan appears normal with high confidence."
        attention_flag = ""
    elif label == "NORMAL":
        explanation = " The scan appears normal, but the model's confidence is low. Consider radiologist review."
        attention_flag = " FLAGGED FOR RADIOLOGIST REVIEW"
    else:
        explanation = " The kidney CT scan shows signs of abnormality. Immediate radiologist attention is recommended."
        attention_flag = " FLAGGED FOR RADIOLOGIST REVIEW"

    return f""" Prediction: {label}
 Confidence: {confidence:.2%}

{explanation}
{attention_flag}"""

# Gradio Interface
demo = gr.Interface(
    fn=predict,
    inputs=gr.Image(type="pil"),
    outputs="text",
    title="Kidney CT Classifier",
    description="Upload a kidney CT image. The model will predict if it's NORMAL or ABNORMAL. Flagged results go to radiologist review."
)

demo.launch()