Real_time_mask_detection - Using MTCNN package for Face detection-Final.py

#!/usr/bin/env python
# coding: utf-8

# # Importing the libraries

# In[1]:


import tensorflow as tf
import numpy as np
import os
import cv2
from mtcnn import MTCNN

from keras_preprocessing import image
from tensorflow.keras.preprocessing import image_dataset_from_directory


# # Loading a custome pre-trained version of MobileNet V2 for image classification

# In[2]:


#Change the working directory to the location within your computer, where the pre-trained MobileNet V2 is saved 
os.chdir('C:\\Users\\Ibrahim Hameem\\Desktop\\Machine Learning\\7. Neural Nets\\Convolutional Neural Network\\Project Face Mask')
print(os.getcwd())


# In[3]:


#Load the pre-trained model
base_model_1 = tf.keras.models.load_model('mask_model_pre-trained_1.h5')


# In[4]:


#We lock the models, such that the imported model is not trainable
base_model_1.trainable = False


# # Core algorithm

# ## Face Detection and Mask or No Mask Classification

# In[5]:


Mask_dict = {'No Mask or Incorrectly masked':1, 'Mask':0}
Color_dict = {1:(0,0,255), 0:(0,255,0)}
prediction_threshold = 0.3


# In[6]:


detector = MTCNN()
video_capture = cv2.VideoCapture(0)

while True:
    _,frame = video_capture.read()
    frame = cv2.resize(frame, (600, 400))
    boxes = detector.detect_faces(frame)
        
    if boxes:
        
        box = boxes[0]['box']
        conf = boxes[0]['confidence']
        x, y, w, h = box[0], box[1], box[2], box[3]
 
        if conf > 0.3:
            roi_color = frame[y-40:y-40+h+80,x-15:x-15+w+30]
            resized = cv2.resize(roi_color,(224,224))
            test_image = image.img_to_array(resized)
            test_image = np.expand_dims(test_image, axis = 0)
            result = base_model_1.predict(test_image)
        
        if result[0][0] >= prediction_threshold:
            prediction = 'No Mask or Incorrectly masked'
        else:
            prediction = 'Mask'
        
        frame = cv2.rectangle(frame, (x-20,y-70), (x-20+w+40, y-70+h+110),Color_dict[Mask_dict[prediction]] ,3)
        frame = cv2.rectangle(frame,(x-110,y+180), (x-110+w+220, y+220),Color_dict[Mask_dict[prediction]],-1)
        frame = cv2.putText(frame,prediction, (x-100, y+210),cv2.FONT_HERSHEY_SIMPLEX,0.5,(255,255,255),2)
        
        if prediction == 'No Mask or Incorrectly masked':
            frame = cv2.putText(frame,str(np.round(result[0][0]*100,2)) + '%', (x+170, y+210),cv2.FONT_HERSHEY_SIMPLEX,0.45,(255,255,255),2)
        else:
            frame =cv2.putText(frame,str(np.round((1-result[0][0])*100,2)) + '%', (x+170, y+210),cv2.FONT_HERSHEY_SIMPLEX,0.45,(255,255,255),2)
        
        cv2.imshow('Video', frame)
         
        
    if cv2.waitKey(1) & 0xFF == ord('q'):
        break
        
video_capture.release()
cv2.destroyAllWindows()


# In[ ]:





# In[ ]: