【656】SegNet 相关说明

 

代码：

?

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87

from keras import Model, layers from keras.layers import Input, Conv2D, BatchNormalization, Activation, Reshape, MaxPooling2D, UpSampling2D   def segnet(pretrained_weights=None, input_size=(512, 512, 3), classNum=2, learning_rate=1e-5):     inputs = Input(input_size)     #encode     #第一层 64,64     conv1 = BatchNormalization()(         Conv2D(64, 3, activation='relu', padding='same', kernel_initializer='he_normal')(inputs))     conv1 = BatchNormalization()(         Conv2D(64, 3, activation='relu', padding='same', kernel_initializer='he_normal')(conv1))     pool1 = MaxPooling2D(pool_size=(2, 2))(conv1)     #第二层 128,128     conv2 = BatchNormalization()(         Conv2D(128, 3, activation='relu', padding='same', kernel_initializer='he_normal')(pool1))     conv2 = BatchNormalization()(         Conv2D(128, 3, activation='relu', padding='same', kernel_initializer='he_normal')(conv2))     pool2 = MaxPooling2D(pool_size=(2, 2))(conv2)     #第三层 256,256,256     conv3 = BatchNormalization()(         Conv2D(256, 3, activation='relu', padding='same', kernel_initializer='he_normal')(pool2))     conv3 = BatchNormalization()(         Conv2D(256, 3, activation='relu', padding='same', kernel_initializer='he_normal')(conv3))     conv3 = BatchNormalization()(         Conv2D(256, 3, activation='relu', padding='same', kernel_initializer='he_normal')(conv3))     pool3 = MaxPooling2D(pool_size=(2, 2))(conv3)     #第四层 512,512,512     conv4 = BatchNormalization()(         Conv2D(512, 3, activation='relu', padding='same', kernel_initializer='he_normal')(pool3))     conv4 = BatchNormalization()(         Conv2D(512, 3, activation='relu', padding='same', kernel_initializer='he_normal')(conv4))     conv4 = BatchNormalization()(         Conv2D(512, 3, activation='relu', padding='same', kernel_initializer='he_normal')(conv4))     pool4 = MaxPooling2D(pool_size=(2, 2))(conv4)     #第五层 512,512,512     conv5 = BatchNormalization()(         Conv2D(512, 3, activation='relu', padding='same', kernel_initializer='he_normal')(pool4))     conv5 = BatchNormalization()(         Conv2D(512, 3, activation='relu', padding='same', kernel_initializer='he_normal')(conv5))     conv5 = BatchNormalization()(         Conv2D(512, 3, activation='relu', padding='same', kernel_initializer='he_normal')(conv5))     pool5 = MaxPooling2D(pool_size=(2, 2))(conv5)     #decode     #上采样     up1 = UpSampling2D(size=(2, 2))(pool5)     #第六层 512,512,512     conv6 = BatchNormalization()(         Conv2D(512, 3, activation='relu', padding='same', kernel_initializer='he_normal')(up1))     conv6 = BatchNormalization()(         Conv2D(512, 3, activation='relu', padding='same', kernel_initializer='he_normal')(conv6))     conv6 = BatchNormalization()(         Conv2D(512, 3, activation='relu', padding='same', kernel_initializer='he_normal')(conv6))     up2 = UpSampling2D(size=(2, 2))(conv6)     #第七层 512,512,512     conv7 = BatchNormalization()(         Conv2D(512, 3, activation='relu', padding='same', kernel_initializer='he_normal')(up2))     conv7 = BatchNormalization()(         Conv2D(512, 3, activation='relu', padding='same', kernel_initializer='he_normal')(conv7))     conv7 = BatchNormalization()(         Conv2D(512, 3, activation='relu', padding='same', kernel_initializer='he_normal')(conv7))     up3 = UpSampling2D(size=(2, 2))(conv7)     #第八层 256,256,256     conv8 = BatchNormalization()(         Conv2D(256, 3, activation='relu', padding='same', kernel_initializer='he_normal')(up3))     conv8 = BatchNormalization()(         Conv2D(256, 3, activation='relu', padding='same', kernel_initializer='he_normal')(conv8))     conv8 = BatchNormalization()(         Conv2D(256, 3, activation='relu', padding='same', kernel_initializer='he_normal')(conv8))     up4 = UpSampling2D(size=(2, 2))(conv8)     # 第八层 256,256,256     conv9 = BatchNormalization()(         Conv2D(128, 3, activation='relu', padding='same', kernel_initializer='he_normal')(up4))     conv9 = BatchNormalization()(         Conv2D(128, 3, activation='relu', padding='same', kernel_initializer='he_normal')(conv9))     up5 = UpSampling2D(size=(2, 2))(conv9)     #第九层 64,64     conv10 = BatchNormalization()(         Conv2D(64, 3, activation='relu', padding='same', kernel_initializer='he_normal')(up5))     conv10 = BatchNormalization()(         Conv2D(64, 3, activation='relu', padding='same', kernel_initializer='he_normal')(conv10))       # softmax输出层     conv11 = Conv2D(1, 1, padding='same', activation='sigmoid')(conv10)       model = Model(inputs=inputs, outputs=conv11)       return model