Коэффициент кости слишком высокий в сегментации изображения с использованием CNN - PullRequest
Новая
       49

Коэффициент кости слишком высокий в сегментации изображения с использованием CNN

0 голосов
/ 26 декабря 2018

Я работаю над проблемой сегментации спутника, используя CNN.Но по достижении 50 эпохи коэффициент костей достиг 0,995.Код для сети приведен ниже. 

def get_unet():
    dropout = 0.20
    axis=3
    inputs = Input((img_rows, img_cols,3 )) # input channels = 3
    conv1 = Conv2D(32, (3, 3), activation='relu', padding='same')(inputs)
    conv1 = BatchNormalization(axis=axis)(conv1)
    conv1 = Conv2D(32, (3, 3), activation='relu', padding='same')(conv1)
    conv1 = BatchNormalization(axis=axis)(conv1)
    conv1 = SpatialDropout2D(dropout)(conv1)
    pool1 = MaxPooling2D(pool_size=(2, 2))(conv1)

    conv2 = Conv2D(64, (3, 3), activation='relu', padding='same')(pool1)
    conv2 = BatchNormalization(axis=axis)(conv2)
    conv2 = Conv2D(64, (3, 3), activation='relu', padding='same')(conv2)
    conv2 = BatchNormalization(axis=axis)(conv2)
    conv2 = SpatialDropout2D(dropout)(conv2)
    pool2 = MaxPooling2D(pool_size=(2, 2))(conv2)

    conv3 = Conv2D(128, (3, 3), activation='relu', padding='same')(pool2)
    conv3 = BatchNormalization(axis=axis)(conv3)
    conv3 = Conv2D(128, (3, 3), activation='relu', padding='same')(conv3)
    conv3 = BatchNormalization(axis=axis)(conv3)
    conv3 = SpatialDropout2D(dropout)(conv3)
    pool3 = MaxPooling2D(pool_size=(2, 2))(conv3)

    conv4 = Conv2D(256, (3, 3), activation='relu', padding='same')(pool3)
    conv4 = BatchNormalization(axis=axis)(conv4)
    conv4 = Conv2D(256, (3, 3), activation='relu', padding='same')(conv4)
    conv4 = BatchNormalization(axis=axis)(conv4)
    conv4 = SpatialDropout2D(dropout)(conv4)
    pool4 = MaxPooling2D(pool_size=(2, 2))(conv4)

    conv5 = Conv2D(512, (3, 3), activation='relu', padding='same')(pool4)
    conv5 = BatchNormalization(axis=axis)(conv5)
    conv5 = Conv2D(512, (3, 3), activation='relu', padding='same')(conv5)
    conv5 = BatchNormalization(axis=axis)(conv5)
    conv5 = SpatialDropout2D(dropout)(conv5)

    up6 = concatenate([Conv2DTranspose(256, (2, 2), strides=(2, 2), padding='same')(conv5), conv4], axis=3)
    conv6 = Conv2D(256, (3, 3), activation='relu', padding='same')(up6)
    conv6 = BatchNormalization(axis=axis)(conv6)
    conv6 = Conv2D(256, (3, 3), activation='relu', padding='same')(conv6)
    conv6 = BatchNormalization(axis=axis)(conv6)
    conv6 = SpatialDropout2D(dropout)(conv6)

    up7 = concatenate([Conv2DTranspose(128, (2, 2), strides=(2, 2), padding='same')(conv6), conv3], axis=3)
    conv7 = Conv2D(128, (3, 3), activation='relu', padding='same')(up7)
    conv7 = BatchNormalization(axis=axis)(conv7)
    conv7 = Conv2D(128, (3, 3), activation='relu', padding='same')(conv7)
    conv7 = BatchNormalization(axis=axis)(conv7)
    conv7 = SpatialDropout2D(dropout)(conv7)

    up8 = concatenate([Conv2DTranspose(64, (2, 2), strides=(2, 2), padding='same')(conv7), conv2], axis=3)
    conv8 = Conv2D(64, (3, 3), activation='relu', padding='same')(up8)
    conv8 = BatchNormalization(axis=axis)(conv8)
    conv8 = Conv2D(64, (3, 3), activation='relu', padding='same')(conv8)
    conv8 = BatchNormalization(axis=axis)(conv8)
    conv8 = SpatialDropout2D(dropout)(conv8)

    up9 = concatenate([Conv2DTranspose(32, (2, 2), strides=(2, 2), padding='same')(conv8), conv1], axis=3)
    conv9 = Conv2D(32, (3, 3), activation='relu', padding='same')(up9)
    conv9 = BatchNormalization(axis=axis)(conv9)
    conv9 = Conv2D(32, (3, 3), activation='relu', padding='same')(conv9)
    conv9 = BatchNormalization(axis=axis)(conv9)
    conv9 = SpatialDropout2D(dropout)(conv9)

    conv10 = Conv2D(1, (1, 1), activation='sigmoid')(conv9)

    model = Model(inputs=[inputs], outputs=[conv10])

    model.compile(optimizer=Adam(lr=1e-5), loss=dice_coef_loss, metrics=[dice_coef])

    return model

Итак, я хочу знать, связана ли она с обучением по сети или с набором данных для обучения / проверки?Есть ли проблема в модели?

...