python+opencv人脸识别（用耶鲁大学的Yale人脸库训练cnn）3

本文链接：https://blog.csdn.net/sereasuesue/article/details/105517839

用耶鲁大学的Yale人脸库，里面包含15个人，每人11张照片，主要包括光照条件的变化，表情的变化，接下来我会把自己的几张照片混进去，看看训练过后能不能被神经网络良好的识别。https://blog.csdn.net/weixin_39329548/article/details/81633120

比这篇修改了一些参数

  def build_model(self, dataset, nb_classes=5):
        # 构建一个空的网络模型，它是一个线性堆叠模型，各神经网络层会被顺序添加，专业名称为序贯模型或线性堆叠模型
        self.model = Sequential()

        self.model.add(Convolution2D(32, 3, 3, border_mode='valid',

                              subsample=(1, 1),

                              dim_ordering='tf',

                              input_shape=(IMAGE_SIZE, IMAGE_SIZE, 3),

                              activation='relu'))

        self.model.add(Convolution2D(32, 3, 3, border_mode='valid',

                              subsample=(1, 1),

                              dim_ordering='tf',

                              activation='relu'))
        self.model.add(MaxPooling2D(pool_size=(2, 2)))
        self.model.add(Dropout(0.20))
        self.model.add(Convolution2D(64, 3, 3, border_mode='valid',

                              subsample=(1, 1),

                              dim_ordering='tf',

                              input_shape=(IMAGE_SIZE, IMAGE_SIZE, 3),

                              activation='relu'))

        self.model.add(Convolution2D(32, 3, 3, border_mode='valid',

                              subsample=(1, 1),

                              dim_ordering='tf',

                              activation='relu'))
        self.model.add(MaxPooling2D(pool_size=(2, 2)))
        self.model.add(Dropout(0.20))
        self.model.add(Flatten())  # 13 Flatten层
        self.model.add(Dense(512,activation = 'relu'))  # 14 Dense层,又被称作全连接层
        self.model.add(Dropout(0.4))  # 16 Dropout层
        self.model.add(Dense(nb_classes, activation = 'softmax'))  # 17 Dense层

        # 输出模型概况
        self.model.summary()

        # 训练模型
    def train(self, dataset, batch_size=15, nb_epoch=120, data_augmentation=True):
            sgd = SGD(lr=0.007, decay=1e-6,
                      momentum=0.9, nesterov=True)  # lr=0.0007 采用SGD+momentum的优化器进行训练，首先生成一个优化器对象
            self.model.compile(loss='categorical_crossentropy',
                               optimizer=sgd,
                               metrics=['accuracy'])  # 完成实际的模型配置工作

_________________________________________________________________
Layer (type) Output Shape Param #
=================================================================
conv2d_1 (Conv2D) (None, 98, 98, 32) 896
_________________________________________________________________
conv2d_2 (Conv2D) (None, 96, 96, 32) 9248
_________________________________________________________________
max_pooling2d_1 (MaxPooling2 (None, 48, 48, 32) 0
_________________________________________________________________
dropout_1 (Dropout) (None, 48, 48, 32) 0
_________________________________________________________________
conv2d_3 (Conv2D) (None, 46, 46, 64) 18496
_________________________________________________________________
conv2d_4 (Conv2D) (None, 44, 44, 64) 36928
_________________________________________________________________
max_pooling2d_2 (MaxPooling2 (None, 22, 22, 64) 0
_________________________________________________________________
dropout_2 (Dropout) (None, 22, 22, 64) 0
_________________________________________________________________
flatten_1 (Flatten) (None, 30976) 0
_________________________________________________________________
dense_1 (Dense) (None, 512) 15860224
_________________________________________________________________
dropout_3 (Dropout) (None, 512) 0
_________________________________________________________________
dense_2 (Dense) (None, 15) 7695
=================================================================
Total params: 15,933,487
Trainable params: 15,933,487
Non-trainable params: 0