Image classification model to detect Face Masks
Consider the dataset that was created with the objective of facilitating the studies of convolutional neural networks and computer vision. Due to the current context related to the coronavirus pandemic, humanity had to adapt to the new reality. The use of face masks has become commonplace in countries around the world. The dataset has 3829 images separated into two distinct categories:
- With mask
- Without mask
Use this dataset to implement an image classification model, which will classify whether the given image is with the mask or without the mask.
import os
import zipfile
import random
import shutil
import tensorflow as tf
from tensorflow.keras.optimizers import RMSprop
from tensorflow.keras.preprocessing.image import ImageDataGenerator
from shutil import copyfile
from os import getcwd
from os import listdir
import cv2
from tensorflow.keras.layers import Conv2D, Input, ZeroPadding2D, BatchNormalization, Activation, MaxPooling2D, Flatten, Dense
from tensorflow.keras.models import Model, load_model
from tensorflow.keras.callbacks import TensorBoard, ModelCheckpoint
from sklearn.model_selection import train_test_split
from sklearn.metrics import f1_score
from sklearn.utils import shuffle
import imutils
import numpy as np
import matplotlib.pyplot as plt
import matplotlib.image as mpimgprint("The number of images with facemask labelling 'with_mask':",len(os.listdir('dataset/train/with_mask')))
print("The number of images with facemask labelling 'without_mask':",len(os.listdir('dataset/train/without_mask/')))The number of images with facemask labelling 'with_mask': 1526
The number of images with facemask labelling 'without_mask': 1538
print("The number of images with facemask in the training set labelled 'yes':", len(os.listdir('dataset/train/with_mask')))
print("The number of images with facemask in the test set labelled 'yes':", len(os.listdir('dataset/test/with_mask')))
print("The number of images without facemask in the training set labelled 'no':", len(os.listdir('dataset/train/without_mask')))
print("The number of images without facemask in the test set labelled 'no':", len(os.listdir('dataset/test/without_mask')))The number of images with facemask in the training set labelled 'yes': 1526
The number of images with facemask in the test set labelled 'yes': 380
The number of images without facemask in the training set labelled 'no': 1538
The number of images without facemask in the test set labelled 'no': 380
model = tf.keras.models.Sequential([
tf.keras.layers.Conv2D(100, (3,3), activation='relu', input_shape=(150, 150, 3)),
tf.keras.layers.MaxPooling2D(2,2),
tf.keras.layers.Conv2D(100, (3,3), activation='relu'),
tf.keras.layers.MaxPooling2D(2,2),
tf.keras.layers.Flatten(),
tf.keras.layers.Dropout(0.5),
tf.keras.layers.Dense(50, activation='relu'),
tf.keras.layers.Dense(2, activation='softmax')
])
model.compile(optimizer='adam', loss='binary_crossentropy', metrics=['acc'])TRAINING_DIR = "dataset/train"
train_datagen = ImageDataGenerator(rescale=1.0/255,
rotation_range=40,
width_shift_range=0.2,
height_shift_range=0.2,
shear_range=0.2,
zoom_range=0.2,
horizontal_flip=True,
fill_mode='nearest')
train_generator = train_datagen.flow_from_directory(TRAINING_DIR,
batch_size=10,
target_size=(150, 150))
VALIDATION_DIR = "dataset/test"
validation_datagen = ImageDataGenerator(rescale=1.0/255)
validation_generator = validation_datagen.flow_from_directory(VALIDATION_DIR,
batch_size=10,
target_size=(150, 150))
checkpoint = ModelCheckpoint('model-{epoch:03d}.model',monitor='val_loss',verbose=0,save_best_only=True,mode='auto')Found 3064 images belonging to 2 classes.
Found 760 images belonging to 2 classes.
history = model.fit_generator(train_generator,
epochs=30,
validation_data=validation_generator,
callbacks=[checkpoint])Epoch 1/30
264/307 [========================>.....] - ETA: 7s - loss: 0.2753 - acc: 0.8967
C:\Users\Pawan\.conda\envs\Deep Learning\lib\site-packages\PIL\Image.py:975: UserWarning: Palette images with Transparency expressed in bytes should be converted to RGBA images
warnings.warn(
307/307 [==============================] - ETA: 0s - loss: 0.2971 - acc: 0.8887INFO:tensorflow:Assets written to: model-001.model\assets
307/307 [==============================] - 61s 198ms/step - loss: 0.2971 - acc: 0.8887 - val_loss: 0.2578 - val_acc: 0.9000
Epoch 2/30
307/307 [==============================] - 59s 193ms/step - loss: 0.2719 - acc: 0.8946 - val_loss: 0.2878 - val_acc: 0.8882
Epoch 3/30
307/307 [==============================] - 61s 197ms/step - loss: 0.2654 - acc: 0.8956 - val_loss: 0.2608 - val_acc: 0.8947
Epoch 4/30
307/307 [==============================] - ETA: 0s - loss: 0.2718 - acc: 0.8949INFO:tensorflow:Assets written to: model-004.model\assets
307/307 [==============================] - 61s 199ms/step - loss: 0.2718 - acc: 0.8949 - val_loss: 0.2487 - val_acc: 0.9053
Epoch 5/30
307/307 [==============================] - 60s 195ms/step - loss: 0.2415 - acc: 0.9093 - val_loss: 0.2749 - val_acc: 0.8974
Epoch 6/30
307/307 [==============================] - ETA: 0s - loss: 0.2376 - acc: 0.9060INFO:tensorflow:Assets written to: model-006.model\assets
307/307 [==============================] - 61s 198ms/step - loss: 0.2376 - acc: 0.9060 - val_loss: 0.2448 - val_acc: 0.9066
Epoch 7/30
307/307 [==============================] - 60s 194ms/step - loss: 0.2348 - acc: 0.9093 - val_loss: 0.6308 - val_acc: 0.8355
Epoch 8/30
307/307 [==============================] - ETA: 0s - loss: 0.2330 - acc: 0.9099INFO:tensorflow:Assets written to: model-008.model\assets
307/307 [==============================] - 61s 198ms/step - loss: 0.2330 - acc: 0.9099 - val_loss: 0.2345 - val_acc: 0.9158
Epoch 9/30
307/307 [==============================] - ETA: 0s - loss: 0.2205 - acc: 0.9138INFO:tensorflow:Assets written to: model-009.model\assets
307/307 [==============================] - 61s 199ms/step - loss: 0.2205 - acc: 0.9138 - val_loss: 0.2290 - val_acc: 0.9105
Epoch 10/30
307/307 [==============================] - 61s 198ms/step - loss: 0.2180 - acc: 0.9151 - val_loss: 0.3110 - val_acc: 0.8789
Epoch 11/30
307/307 [==============================] - ETA: 0s - loss: 0.2053 - acc: 0.9197INFO:tensorflow:Assets written to: model-011.model\assets
307/307 [==============================] - 60s 196ms/step - loss: 0.2053 - acc: 0.9197 - val_loss: 0.2205 - val_acc: 0.9263
Epoch 12/30
307/307 [==============================] - 59s 192ms/step - loss: 0.2131 - acc: 0.9227 - val_loss: 0.2603 - val_acc: 0.9118
Epoch 13/30
307/307 [==============================] - 59s 192ms/step - loss: 0.2134 - acc: 0.9223 - val_loss: 0.2211 - val_acc: 0.9079
Epoch 14/30
307/307 [==============================] - 59s 192ms/step - loss: 0.2095 - acc: 0.9220 - val_loss: 0.2217 - val_acc: 0.9250
Epoch 15/30
307/307 [==============================] - 59s 192ms/step - loss: 0.1989 - acc: 0.9253 - val_loss: 0.2337 - val_acc: 0.9171
Epoch 16/30
307/307 [==============================] - ETA: 0s - loss: 0.1988 - acc: 0.9181INFO:tensorflow:Assets written to: model-016.model\assets
307/307 [==============================] - 60s 197ms/step - loss: 0.1988 - acc: 0.9181 - val_loss: 0.2117 - val_acc: 0.9382
Epoch 17/30
307/307 [==============================] - ETA: 0s - loss: 0.1924 - acc: 0.9240INFO:tensorflow:Assets written to: model-017.model\assets
307/307 [==============================] - 60s 196ms/step - loss: 0.1924 - acc: 0.9240 - val_loss: 0.2101 - val_acc: 0.9263
Epoch 18/30
307/307 [==============================] - 59s 192ms/step - loss: 0.1828 - acc: 0.9315 - val_loss: 0.2617 - val_acc: 0.9224
Epoch 19/30
307/307 [==============================] - 59s 192ms/step - loss: 0.1937 - acc: 0.9298 - val_loss: 0.2141 - val_acc: 0.9368
Epoch 20/30
307/307 [==============================] - 59s 192ms/step - loss: 0.1756 - acc: 0.9341 - val_loss: 0.2909 - val_acc: 0.9145
Epoch 21/30
307/307 [==============================] - 59s 192ms/step - loss: 0.1792 - acc: 0.9341 - val_loss: 0.2448 - val_acc: 0.9303
Epoch 22/30
307/307 [==============================] - 59s 191ms/step - loss: 0.1669 - acc: 0.9337 - val_loss: 0.2193 - val_acc: 0.9289
Epoch 23/30
307/307 [==============================] - ETA: 0s - loss: 0.1848 - acc: 0.9324INFO:tensorflow:Assets written to: model-023.model\assets
307/307 [==============================] - 60s 196ms/step - loss: 0.1848 - acc: 0.9324 - val_loss: 0.1987 - val_acc: 0.9368
Epoch 24/30
307/307 [==============================] - ETA: 0s - loss: 0.1637 - acc: 0.9373INFO:tensorflow:Assets written to: model-024.model\assets
307/307 [==============================] - 60s 196ms/step - loss: 0.1637 - acc: 0.9373 - val_loss: 0.1807 - val_acc: 0.9408
Epoch 25/30
307/307 [==============================] - 59s 192ms/step - loss: 0.1732 - acc: 0.9370 - val_loss: 0.2141 - val_acc: 0.9513
Epoch 26/30
307/307 [==============================] - 59s 192ms/step - loss: 0.1722 - acc: 0.9344 - val_loss: 0.2180 - val_acc: 0.9395
Epoch 27/30
307/307 [==============================] - 59s 192ms/step - loss: 0.1730 - acc: 0.9354 - val_loss: 0.2360 - val_acc: 0.9382
Epoch 28/30
307/307 [==============================] - 59s 192ms/step - loss: 0.1679 - acc: 0.9354 - val_loss: 0.2742 - val_acc: 0.9250
Epoch 29/30
307/307 [==============================] - 59s 191ms/step - loss: 0.1590 - acc: 0.9360 - val_loss: 0.2110 - val_acc: 0.9342
Epoch 30/30
307/307 [==============================] - 59s 191ms/step - loss: 0.1487 - acc: 0.9409 - val_loss: 0.2508 - val_acc: 0.9342
face_clsfr=cv2.CascadeClassifier('haarcascade_frontalface_default.xml')model.save_weights("model.h5")N = 30
plt.style.use("ggplot")
plt.figure()
plt.plot(np.arange(0, N), history.history["loss"], label="train_loss")
plt.plot(np.arange(0, N), history.history["val_loss"], label="val_loss")
plt.plot(np.arange(0, N), history.history["acc"], label="train_acc")
plt.plot(np.arange(0, N), history.history["val_acc"], label="val_acc")
plt.title("Training Loss and Accuracy")
plt.xlabel("Epoch #")
plt.ylabel("Loss/Accuracy")
plt.legend(loc="lower left")<matplotlib.legend.Legend at 0x203b5f80c70>
labels_dict={0:'without_mask',1:'with_mask'}
color_dict={0:(0,0,255),1:(0,255,0)}
size = 4
webcam = cv2.VideoCapture(0) #Use camera 0
# We load the xml file
classifier = cv2.CascadeClassifier('haarcascade_frontalface_default.xml')
test_directory = 'tests'
for root, directories, file in os.walk(test_directory):
files = file
for file in files:
im = cv2.imread(os.path.join(test_directory , file))
# Resize the image to speed up detection
mini = cv2.resize(im, (im.shape[1] // size, im.shape[0] // size))
# detect MultiScale / faces
faces = classifier.detectMultiScale(mini)
# Draw rectangles around each face
for f in faces:
(x, y, w, h) = [v * size for v in f] #Scale the shapesize backup
#Save just the rectangle faces in SubRecFaces
face_img = im[y:y+h, x:x+w]
resized=cv2.resize(face_img,(150,150))
normalized=resized/255.0
reshaped=np.reshape(normalized,(1,150,150,3))
reshaped = np.vstack([reshaped])
result=model.predict(reshaped)
#print(result)
label=np.argmax(result,axis=1)[0]
cv2.rectangle(im,(x,y),(x+w,y+h),color_dict[label],2)
cv2.rectangle(im,(x,y-40),(x+w,y),color_dict[label],-1)
cv2.putText(im, labels_dict[label], (x, y-10),cv2.FONT_HERSHEY_SIMPLEX,0.8,(255,255,255),2)
cv2.imwrite(os.path.join(os.getcwd()+'output')+str('output_'+file+'.jpg'), im)
import glob
imdir = "output"
ext = ['jpg']
files = []
[files.extend(glob.glob(imdir + '*.' + e)) for e in ext]
images = [cv2.imread(file) for file in files]
for image in images:
image = cv2.cvtColor(image, cv2.COLOR_BGR2RGB)
plt.imshow(image)
plt.show()
