vedio.py

import os
import tensorflow as tf
import numpy as np
import cv2
import time
from matplotlib import pyplot as plt
from filter import apply

# supress tensorflow warning
os.environ['TF_CPP_MIN_LOG_LEVEL'] = '3'
# from tensorflow.keras.models import load_model

# load model
# model1 is better than model
model = tf.keras.models.load_model('model/model1.h5', compile=False)
# model = tf.keras.models.load_model('model/model.h5',compile = False)


def detect_points(face_img):
    face_array = np.array(face_img)
    x_test = np.expand_dims(face_array, axis=0)
    x_test = np.expand_dims(x_test, axis=3)
    y_test = model.predict(x_test)
    return y_test


# Load haarcascade
face_cascade = cv2.CascadeClassifier('haarcascade_frontalface_default.xml')
dimensions = (96, 96)

# declare vid object
vid = cv2.VideoCapture(0)
while True:
    # Process the frame
    ret, frame = vid.read()
    img = frame  # 480 640 3
    gray_img = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
    faces = face_cascade.detectMultiScale(gray_img, 1.3, 5)

    # No face found
    if len(faces) == 0:
        print("No faces found!")

    start = time.time()
    for i, (x, y, w, h) in enumerate(faces):
        # isolate the face
        just_face = cv2.resize(gray_img[y:y+h, x:x+w], dimensions)
        just_color_face = cv2.resize(img[y:y+h, x:x+w], dimensions)

        # get landmarks
        label_point = detect_points(just_face)

        # apply all the filters you want
        types = "glasses and moustache"
        just_color_face = apply(just_color_face,label_point,types)

        # resize and fit back the isolated face in original image
        just_color_face = cv2.resize(just_color_face, (h, w))
        img[y:y+h, x:x+w] = just_color_face

        # Display the resulting frame
        cv2.imshow('frame', img)

        endtime = time.time()
        ## it takes 0.1 second
        print("Processed Frame in :",endtime - start)

        # the 'q' button is set as the
        # quitting button you may use any
        # desired button of your choice
        if cv2.waitKey(1) & 0xFF == ord('q'):
            break