Python-⼈脸识别检测是否佩戴⼝罩使⽤⼝罩数据集
本博客运⾏环境为jupyter下python3.6
完成对⼝罩佩戴与否的模型训练，采取合适的特征提取⽅法，输出模型训练精度和测试精度（F1-score和ROC）；完成⼀个摄像头采集⾃⼰⼈脸、并能实时分类判读（输出分类⽂字）的程序。
⼝罩数据集
链接:
提取码: fv15
图⽚预处理
把数据集中的图⽚⼈脸部分裁剪下来。记得修改路径为⾃⼰的路径哦。
import dlib        # ⼈脸识别的库dlib
python怎么读取dat文件import numpy as np  # 数据处理的库numpy
import cv2          # 图像处理的库OpenCv
import os
# dlib预测器
detector = _frontal_face_detector()
predictor = dlib.shape_predictor('shape_predictor_68_face_landmarks.dat')
# 读取图像的路径
path_read ="data"
for file_name in os.listdir(path_read):
#aa是图⽚的全路径
aa=(path_read +"/"+file_name)
#读⼊的图⽚的路径中含⾮英⽂
img=cv2.imdecode(np.fromfile(aa, dtype=np.uint8), cv2.IMREAD_UNCHANGED) #获取图⽚的宽⾼
img_shape=img.shape
img_height=img_shape[0]
img_width=img_shape[1]
# ⽤来存储⽣成的单张⼈脸的路径
path_save="maskdata"
# dlib检测
dets = detector(img,1)
print("⼈脸数：",len(dets))
for k, d in enumerate(dets):
if len(dets)>1:
continue
# 计算矩形⼤⼩
# (x,y), (宽度width, ⾼度height)
pos_start =tuple([d.left(), d.top()])
pos_end =tuple([d.right(), d.bottom()])
# 计算矩形框⼤⼩
height = d.bottom()-d.top()
width = d.right()-d.left()
# 根据⼈脸⼤⼩⽣成空的图像
img_blank = np.zeros((height, width,3), np.uint8)
for i in range(height):
p()+i>=img_height:# 防⽌越界
continue
for j in range(width):
if d.left()+j>=img_width:# 防⽌越界
continue
img_blank[i][j]= p()+i][d.left()+j]
img_blank = size(img_blank,(200,200), interpolation=cv2.INTER_CUBIC)        cv2.imencode('.jpg', img_blank)[1].tofile(path_save+"/"+file_name)# 正确⽅法导⼊数据集
划分完成后，导⼊数据集。
代码如下：
import keras
import os, shutil
train_havemask_dir="maskdata/train/mask/"
train_nomask_dir="maskdata/train/nomask/"
test_havemask_dir="maskdata/test/mask/"
test_nomask_dir="maskdata/test/nomask/" validation_havemask_dir="maskdata/validation/mask/" validation_nomask_dir="maskdata/validation/nomask/" train_dir="maskdata/train/"
test_dir="maskdata/test/"
validation_dir="maskdata/validation/"
创建模型
代码如下：
#创建模型
from keras import layers
from keras import models
model = models.Sequential()
model.add(layers.Conv2D(32,(3,3), activation='relu',                        input_shape=(150,150,3)))
model.add(layers.MaxPooling2D((2,2)))
model.add(layers.Conv2D(64,(3,3), activation='relu')) model.add(layers.MaxPooling2D((2,2)))
model.add(layers.Conv2D(128,(3,3), activation='relu')) model.add(layers.MaxPooling2D((2,2)))
model.add(layers.Conv2D(128,(3,3), activation='relu')) model.add(layers.MaxPooling2D((2,2)))
model.add(layers.Flatten())
model.add(layers.Dense(512, activation='relu')) model.add(layers.Dense(1, activation='sigmoid'))
查看模型：
model.summary()
运⾏结果如下：
_________________________________________________________________
Layer (type) Output Shape Param #
================================================================= conv2d_1 (Conv2D) (None, 148, 148, 32) 896
_________________________________________________________________
max_pooling2d_1 (MaxPooling2 (None, 74, 74, 32) 0
_________________________________________________________________
conv2d_2 (Conv2D) (None, 72, 72, 64) 18496
_________________________________________________________________
max_pooling2d_2 (MaxPooling2 (None, 36, 36, 64) 0
_________________________________________________________________
conv2d_3 (Conv2D) (None, 34, 34, 128) 73856
_________________________________________________________________
max_pooling2d_3 (MaxPooling2 (None, 17, 17, 128) 0
_________________________________________________________________
conv2d_4 (Conv2D) (None, 15, 15, 128) 147584
_________________________________________________________________
max_pooling2d_4 (MaxPooling2 (None, 7, 7, 128) 0
_________________________________________________________________
flatten_1 (Flatten) (None, 6272) 0
_________________________________________________________________
dense_1 (Dense) (None, 512) 3211776
_________________________________________________________________
dense_2 (Dense) (None, 1) 513
================================================================= Total params: 3,453,121
Trainable params: 3,453,121
Non-trainable params: 0
_________________________________________________________________
归⼀化处理
代码如下：
from keras import optimizers
modelpile(loss='binary_crossentropy',
optimizer=optimizers.RMSprop(lr=1e-4),
metrics=['acc'])
from keras.preprocessing.image import ImageDataGenerator
# All images will be rescaled by 1./255
train_datagen = ImageDataGenerator(rescale=1./255)
validation_datagen=ImageDataGenerator(rescale=1./255)
test_datagen = ImageDataGenerator(rescale=1./255)
train_generator = train_datagen.flow_from_directory(
# ⽬标⽂件⽬录
train_dir,
#所有图⽚的size必须是150x150
target_size=(150,150),
batch_size=20,
# Since we use binary_crossentropy loss, we need binary labels
class_mode='binary')
validation_generator = test_datagen.flow_from_directory(
validation_dir,
target_size=(150,150),
batch_size=20,
class_mode='binary')
test_generator = test_datagen.flow_from_directory(test_dir,
target_size=(150,150),
batch_size=20,
class_mode='binary')
for data_batch, labels_batch in train_generator:
print('data batch shape:', data_batch.shape)
print('labels batch shape:', labels_batch)
break
运⾏结果如下：
data batch shape: (20, 150, 150, 3)
labels batch shape: [1. 0. 1. 1. 0. 0. 0. 1. 1. 1. 0. 1. 0. 1. 1. 1. 1. 1. 1. 1.]训练模型
代码如下：
#耗时长
history = model.fit_generator(
train_generator,
steps_per_epoch=100,
epochs=10,
validation_data=validation_generator,
validation_steps=50)

版权声明：本站内容均来自互联网，仅供演示用，请勿用于商业和其他非法用途。如果侵犯了您的权益请与我们联系QQ:729038198，我们将在24小时内删除。