import gradio as gr
import tensorflow as tf
from tensorflow.keras.applications.mobilenet_v2 import preprocess_input
from tensorflow.keras.models import load_model
from tensorflow.keras.preprocessing import image
import numpy as np
from PIL import Image


# Load the model
model = load_model("keras_model.h5", compile=False)

# Load the labels
class_names = open("labels.txt", "r").readlines()


def predict_image(img):
    # Resize the image
    img = Image.fromarray(img.astype('uint8'), 'RGB')
    img = img.resize((224, 224), resample=Image.BILINEAR)

    # Preprocess the image
    img_array = image.img_to_array(img)
    img_array = preprocess_input(img_array)

    # Expand the dimensions to create a batch of size 1
    img_batch = tf.expand_dims(img_array, axis=0)

    # Predict the class probabilities
    preds = model.predict(img_batch)
    class_idx = tf.argmax(preds, axis=1)[0]
    class_name = class_names[class_idx].strip()
    confidence_score = float(preds[0][class_idx])  # convert to float
    
    if class_idx == 5:
        return "We couldn't detect anything from this image. Please try with a different image."
    elif confidence_score >= 0.70:
        return f"There is a {confidence_score*100:.2f}% chance for this image to be {class_name}. Even though it has good accuracy, please consult a doctor for confirmation."
    elif 0.50 <= confidence_score < 0.70:
        return f"There is a {confidence_score*100:.2f}% chance for this image to be {class_name}, but considering the accuracy, it's better to consult a doctor before using our service."
    else:
        return f"There is a {confidence_score*100:.2f}% chance for this image to be {class_name}. Since the accuracy is very low, please consider a doctor's advice and we recommend you not to rely on our predictions."


# Launch the Gradio interface
iface = gr.Interface(fn=predict_image, inputs="image", outputs="text", title="Bee4Med - Skin Disease Classifier",
                     description="This is a machine learning model that predicts skin disease from an image(limited dataset). However, please note that there are chances that the predictions may go wrong, and we strongly recommend you to consult a doctor for confirmation.")

# Launch the interface
iface.launch()