ValueError：“ conv2d_2 / convolution”的2中减去3导致的负尺寸大小

时间：2020-01-29 05:14:10

标签： python opencv keras conv-neural-network 

# import the necessary packages
from keras.initializers import glorot_uniform
from keras.layers import AveragePooling2D, Input, Add
from keras.models import Model
from keras.layers.normalization import BatchNormalization
from keras.layers.convolutional import Conv2D
from keras.layers.convolutional import MaxPooling2D
from keras.layers.core import Activation
from keras.layers.core import Flatten
from keras.layers.core import Dropout
from keras.layers.core import Dense


class SmallerVGGNet:
    @staticmethod
    def build(width, height, depth, classes, finalact):

        X1 = Input(shape=(height, width, depth))
        output = -1

        # # CONV => RELU => POOL
        X = Conv2D(16, (3, 3), padding="same", strides=(1, 1), name="con_layer1")(X1)
        X = BatchNormalization(axis=output)(X)
        X = Activation("relu")(X)
        X = MaxPooling2D(pool_size=(3, 3))(X)

        X = Conv2D(32, (3, 3), padding="same", strides=(2, 2), name="con_layer2")(X)
        X = BatchNormalization(axis=output)(X)
        X = Activation("relu")(X)
        X = Conv2D(32, (3, 3), padding="same", strides=(1, 1), name="con_layer3")(X)
        X = Activation("relu")(X)
        X = BatchNormalization(axis=output)(X)
        X = MaxPooling2D(pool_size=(3, 3))(X)

        # First component
        X0 = Conv2D(256, (3, 3), strides=(1, 1), padding='same', kernel_initializer=glorot_uniform(seed=0))(X)
        X0 = BatchNormalization(axis=3)(X0)
        X0 = Activation("relu")(X0)

        # (CONV => RELU) * 2 => POOL
        X = Conv2D(64, (3, 3), padding="same", strides=(2, 2), name="con_layer4")(X0)
        X = BatchNormalization(axis=output)(X)
        X = Activation("relu")(X)
        X = Conv2D(64, (3, 3), padding="same", strides=(1, 1), name="con_layer5")(X)
        X = BatchNormalization(axis=output)(X)
        X = Activation("relu")(X)
        X = AveragePooling2D(pool_size=(2, 2))(X)

        # Second Component
        X0 = Conv2D(512, (3, 3), strides=(2, 2), padding='valid', kernel_initializer=glorot_uniform(seed=0))(X)
        X0 = BatchNormalization(axis=3)(X0)
        X0 = Activation("relu")(X0)

        # (CONV => RELU) * 2 => POOL
        X = Conv2D(128, (3, 3), padding="same", strides=(1, 1), name="con_layer6")(X1)
        X = BatchNormalization(axis=3)(X)
        X = Activation("relu")(X)
        X = Conv2D(128, (3, 3), padding="same", strides=(2, 2), name="con_layer7")(X)
        X = BatchNormalization(axis=output)(X)
        X = Activation("relu")(X)
        X = MaxPooling2D(pool_size=(2, 2))(X)

        # Third Component
        X0 = Conv2D(1024, (3, 3), strides=(2, 2), padding='valid', kernel_initializer=glorot_uniform(seed=0))(X1)
        X0 = BatchNormalization(axis=3)(X0)
        X0 = Dense(128, activation="relu")(X0)
        X0 = Activation("relu")(X0)

        X = Flatten()(X1)
        X = Dense(128)(X)
        X = BatchNormalization()(X)
        X = Dropout(0.5)(X)
        output = Dense(classes, activation=finalact)(X)

        model = Model(inputs=[X1], outputs=output)

        print(model.summary())
        return model

ValueError：负尺寸大小是由于输入形状为[？，2,2,64]，[3,3,64,512]的'conv2d_2 / convolution'（op：'Conv2D'）从2中减去3引起的。 / p>

该错误是第50行的原因。当我将值更改为（X）时，将发生此错误。如果我将值更改为（X1），它将运行，但我需要传递值（X）

1 个答案:

答案 0 :(得分：0)

最简单的解决方法是将padding='same'添加到所有池化层。这是可以解决负尺寸问题的更新代码。 

# import the necessary packages
from keras.initializers import glorot_uniform
from keras.layers import AveragePooling2D, Input, Add
from keras.models import Model
from keras.layers.normalization import BatchNormalization
from keras.layers.convolutional import Conv2D
from keras.layers.convolutional import MaxPooling2D
from keras.layers.core import Activation
from keras.layers.core import Flatten
from keras.layers.core import Dropout
from keras.layers.core import Dense


class SmallerVGGNet:
    @staticmethod
    def build(width, height, depth, classes, finalact):

        X1 = Input(shape=(height, width, depth))
        output = -1

        # # CONV => RELU => POOL
        X = Conv2D(16, (3, 3), padding='same', strides=(1, 1), name="con_layer1")(X1)
        X = BatchNormalization(axis=output)(X)
        X = Activation("relu")(X)
        X = MaxPooling2D(pool_size=(3, 3), padding='same')(X)

        X = Conv2D(32, (3, 3), padding='same', strides=(2, 2), name="con_layer2")(X)
        X = BatchNormalization(axis=output)(X)
        X = Activation("relu")(X)
        X = Conv2D(32, (3, 3), padding='same', strides=(1, 1), name="con_layer3")(X)
        X = Activation("relu")(X)
        X = BatchNormalization(axis=output)(X)
        X = MaxPooling2D(pool_size=(3, 3), padding='same')(X)

        # First component
        X0 = Conv2D(256, (3, 3), strides=(1, 1), padding='same', kernel_initializer=glorot_uniform(seed=0))(X)
        X0 = BatchNormalization(axis=3)(X0)
        X0 = Activation("relu")(X0)

        # (CONV => RELU) * 2 => POOL
        X = Conv2D(64, (3, 3), padding='same', strides=(2, 2), name="con_layer4")(X0)
        X = BatchNormalization(axis=output)(X)
        X = Activation("relu")(X)
        X = Conv2D(64, (3, 3), padding='same', strides=(1, 1), name="con_layer5")(X)
        X = BatchNormalization(axis=output)(X)
        X = Activation("relu")(X)
        X = AveragePooling2D(pool_size=(2, 2),padding='same')(X)

        # Second Component
        X0 = Conv2D(512, (3, 3), strides=(2, 2), padding='valid', kernel_initializer=glorot_uniform(seed=0))(X)
        X0 = BatchNormalization(axis=3)(X0)
        X0 = Activation("relu")(X0)

        # (CONV => RELU) * 2 => POOL
        X = Conv2D(128, (3, 3), padding='same', strides=(1, 1), name="con_layer6")(X1)
        X = BatchNormalization(axis=3)(X)
        X = Activation("relu")(X)
        X = Conv2D(128, (3, 3), padding='same', strides=(2, 2), name="con_layer7")(X)
        X = BatchNormalization(axis=output)(X)
        X = Activation("relu")(X)
        X = MaxPooling2D(pool_size=(2, 2),padding='same')(X)

        # Third Component
        X0 = Conv2D(1024, (3, 3), strides=(2, 2), padding='valid', kernel_initializer=glorot_uniform(seed=0))(X1)
        X0 = BatchNormalization(axis=3)(X0)
        X0 = Dense(128, activation="relu")(X0)
        X0 = Activation("relu")(X0)

        X = Flatten()(X1)
        X = Dense(128)(X)
        X = BatchNormalization()(X)
        X = Dropout(0.5)(X)
        output = Dense(classes, activation=finalact)(X)

        model = Model(inputs=[X1], outputs=output)

        print(model.summary())
        return model
相关问题
最新问题