# -*- coding: utf-8 -*-
"""app

Automatically generated by Colaboratory.

Original file is located at
    https://colab.research.google.com/drive/1bK9oazNpp0ELz8EATsAmWYgKaiUpdV3W
"""

import gradio as gr

import tensorflow as tf
import keras
import numpy as np
import gradio
import requests
from tensorflow.keras.models import load_model
import json
from os.path import dirname, realpath, join

model=load_model("model.h5")

with open("label.json") as labels_file:
    labels = json.load(labels_file)

def trans_image(pillow_image):
    a = np.array(pillow_image.resize((128,128)))
    a = a.reshape(1,128,128,3)
    return a

def brain(predict):
    if predict==0:
        return 'Glioma Tumor'
    elif predict==1:
        return 'Meningioma Tumor'
    elif predict==2:
        return 'No Tumor or Normal'
    else:
        return 'Pituitary Tumor'
    
def predictions(x):
    a = model.predict(x)
    classi = np.where(a == np.amax(a))[1][0]
    return str(a[0][classi]*100), brain(classi)

def predicts(img):
    img = img.reshape((1, 128, 128, 3))
    prediction = model.predict(img).flatten()

    return {labels[i]: float(prediction[i]) for i in range(4)}
image = gradio.inputs.Image(shape=(128, 128), source="upload")
label = gr.outputs.Label(num_top_classes= 5)

iface = gr.Interface(
    fn = predicts, 
    inputs = image, 
    outputs = label,
    )
iface.launch(debug=True, share=False)