基于Keras的人脸识别

第1关：数据集加载与预处理

本关主要带大家掌握如何进行数据的加载和预处理。

import warnings
warnings.filterwarnings("ignore")
import glob
import cv2
import numpy as np
import keras
from sklearn.model_selection import train_test_split

# 人物的Label-encoding，构造字典 map_characters 存储 label
# ********** Begin *********#
map_characters = {0:'A',1:'C',2:'D',3:'F',4:'G',5:'H',6:'I',7:'J',8:'K',9:'L', 10:'M',11:'P',12:'R',13:'S', 14:'T', 15:'V' }
# ********** End *********#
imgsPath = "/opt/train"
# 图片裁切大小
img_width = 42
img_height = 42
# 验证集比例
test_size = 0.15
num_classes = len(map_characters) # 要辨识的角色个数
# 加载图片
def load_pictures():
    pics = []
    labels = []
# ********** Begin *********#
    for key, v in map_characters.items():
        pictures = [k for k in glob.glob(imgsPath + "/" + v + "/*")]
        for pic in pictures:
            tmp_img = cv2.imread(pic)
            tmp_img = cv2.cvtColor(tmp_img, cv2.COLOR_BGR2RGB)
            tmp_img = cv2.resize(tmp_img, (img_width, img_height))
            pics.append(tmp_img)
            labels.append(key)
# ********** End *********#
    return np.array(pics), np.array(labels)
# 取得训练集与验证集
def get_dataset():
# ********** Begin *********#
    X, Y = load_pictures()
    Y = keras.utils.to_categorical(Y, num_classes) 
    X_train, X_test, y_train, y_test = train_test_split(X, Y, test_size = 0.15)

# ********** End *********#
   
    return X_train, X_test, y_train, y_test
X_train, X_test, y_train, y_test = get_dataset()

第2关：构建网络模型

本关主要带大家掌握如何构建网络模型。

import warnings
warnings.filterwarnings("ignore")
from keras.models import Sequential
from keras.layers import Conv2D, MaxPooling2D
from keras.layers import Dense, Dropout, Activation, Flatten

img_width = 42
img_height = 42

map_characters = {0: 'A', 1: 'C', 2: 'D',
        3: 'F', 4: 'G', 5: 'H', 6: 'I',
        7: 'J', 8: 'K', 9: 'L', 10:'M',
        11:'P', 12:'R', 13:'S', 14:'T', 15:'V'}

num_classes = len(map_characters) # 要辨识的角色个数


# 定义模型
# input_shape 由函数调用时传入，为（42，42，3）
def create_model_six_conv(input_shape):
    # ********** Begin *********#
    model = Sequential()
    model.add(Conv2D(32, (3, 3), padding='same', activation='relu', input_shape=input_shape))
    model.add(Conv2D(32, (3, 3), activation='relu'))
    model.add(MaxPooling2D(pool_size=(2, 2)))
    model.add(Dropout(0.2))
    model.add(Conv2D(64, (3, 3), padding='same', activation='relu'))
    model.add(Conv2D(64, (3, 3), activation='relu'))
    model.add(MaxPooling2D(pool_size=(2, 2)))
    model.add(Dropout(0.2))
    model.add(Conv2D(128, (3, 3), padding='same', activation='relu'))
    model.add(Conv2D(128, (3, 3), activation='relu'))
    model.add(MaxPooling2D(pool_size=(2, 2)))
    model.add(Dropout(0.2))
    model.add(Flatten())
    model.add(Dense(1024, activation='relu'))
    model.add(Dropout(0.5))
    model.add(Dense(num_classes, activation='softmax'))


    # ********** End **********#

    return model;

#圖像的shape是 (42,42,3)
model = create_model_six_conv((img_height, img_width, 3)) # 初始化一個模型

第3关：训练模型

本关主要带大家掌握如何训练网络模型。

import warnings  
warnings.filterwarnings("ignore")  
import os  
from keras.optimizers import SGD  
from keras.callbacks import LearningRateScheduler, ModelCheckpoint
os.environ['TF_CPP_MIN_LOG_LEVEL'] = '3'
# Model Compile  
# ********** Begin *********#  
lr = 0.01  
sgd = SGD(lr=lr, decay=1e-6, momentum=0.9, nesterov=True)  
model.compile(loss='categorical_crossentropy',  
         optimizer=sgd,  
         metrics=['accuracy'])  
# ********** End *********#
# Training  
# ********** Begin *********#  
def lr_schedule(epoch):  
    return lr*(0.1**int(epoch/10))  
# ********** End *********#
# ********** Begin *********#  
batch_size = 32  
epochs = 30  
# ********** End *********#  
filepath = "model.h5"  
# ********** Begin *********#  
# 请不要删除参数 verbose，否则会影响评测  
history = model.fit(X_train, y_train,  
    batch_size=batch_size,  
    epochs=epochs,  
    validation_data=(X_test, y_test),  
    shuffle=True,  
    verbose = 0,  
    callbacks=[LearningRateScheduler(lr_schedule),  
        ModelCheckpoint(filepath, save_best_only=True)])  
# ********** End *********#

第4关：验证评估

本关主要带大家掌握如何对训练过的模型进行评估。

import warnings  
warnings.filterwarnings("ignore")  
import glob  
import cv2  
import numpy as np  
import keras  
import os  
import pandas as pd  
from keras.models import load_model  
import os
# 屏蔽系统信息  
os.environ['TF_CPP_MIN_LOG_LEVEL'] = '3'
def load_test_set(path):  
    pics, labels = [], []
    map_characters = {0: 'A', 1: 'C', 2: 'D',  
        3: 'F', 4: 'G', 5: 'H', 6: 'I',  
        7: 'J', 8: 'K', 9: 'L', 10:'M',  
        11:'P', 12:'R', 13:'S', 14:'T', 15:'V'}
    num_classes = len(map_characters)
    img_width = 42  
    img_height = 42
    # ********** Begin *********#  
    # 一定要反转字典  
    reverse_dict = {v:k for k,v in map_characters.items()}
    # 获取目录 "/opt/test/" 下的所有内容（所有图片的地址+图片名）  
    for pic in glob.glob(path+'*.*'):
        # os.path.basename(pic):返回每张图片的名字  
        char_name = "".join(os.path.basename(pic).split('_')[0])
        # 查找 人名 是否在 lable-value 里面  
        if char_name in reverse_dict:  
            temp = cv2.imread(pic)  
            temp = cv2.cvtColor(temp, cv2.COLOR_BGR2RGB)  
            temp = cv2.resize(temp, (img_height,img_width)).astype('float32') / 255.  
            pics.append(temp)  
            labels.append(reverse_dict[char_name])  # 得到的将是 int  
    X_test = np.array(pics)  
    y_test = np.array(labels)  
    y_test = keras.utils.to_categorical(y_test, num_classes) # one-hot编码  
    # ********** End *********#
    return X_test, y_test
imgsPath = "/opt/test/"
X_valtest, y_valtest = load_test_set(imgsPath)
# 预测对比  
def acc():  
    # 把训练时 val_loss 最小的模型载入  
    model = load_model("model.h5")
    # 预测与对比  
    # ********** Begin *********#  
    y_pred = model.predict_classes(X_valtest)  
    acc = np.sum(y_pred==np.argmax(y_valtest, axis=1))/np.size(y_pred)  
    # ********** End *********#  
    return(acc)