import gradio as gr
import cv2
import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
from tensorflow.keras.models import load_model

facial = load_model('facial_recognition_model.h5')


def facial_prediction(img):
    img = cv2.resize(img, (90, 90))
    labels = {0: 'Angry', 1: 'Fear', 2: 'Happy',
              3: 'Neutral', 4: 'Sad', 5: 'Suprise'}
    color = {0: (0, 0, 255), 1: (1, 1, 1), 2: (0, 255, 0), 3: (
        255, 255, 255), 4: (100, 100, 100), 5: (255, 255, 255)}
    x = []
    x.append(img)
    x = np.array(x)
    pred = facial.predict(x)
    pred = np.argmax(pred)
    return labels[pred], color[pred]

face_classifier = cv2.CascadeClassifier(
    cv2.data.haarcascades + "haarcascade_frontalface_default.xml")


def classify_image(img):
    gray_img = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
    faces_index = face_classifier.detectMultiScale(gray_img, 1.0485258, 6)
    for (x, y, w, h) in faces_index:
        face = img[y:y+h, x:x+w]
        face = cv2.resize(face, (90, 90))
        text, color = facial_prediction(face)
        cv2.rectangle(img, (x, y), (x+w, y+h), color, thickness=4)
        cv2.putText(img, text, (x, y-10),
                    cv2.FONT_HERSHEY_SIMPLEX, 1, color, 2)
    return img


gr.Interface(fn=classify_image,
             inputs=gr.Image(),
             outputs=gr.Image(),
             examples=["sample_image1.jpeg"]
             ).launch()