mymodel.py

from keras.layers import Input, Dense, Conv2D, BatchNormalization,Activation,Flatten,MaxPooling2D,Dropout
from keras.layers import concatenate
from keras.models import Model

def model_build(width,height,depth,filters=128):
    inputs = Input((width,height,depth))
    conv1 = Conv2D(filters,(2,1),kernel_initializer="he_uniform",padding="same")(inputs)
    conv2 = Conv2D(filters,(1,2),kernel_initializer="he_uniform",padding="same")(inputs)
    conv11 = Conv2D(filters,(2,1),kernel_initializer="he_uniform",padding="same")(conv1)
    conv12 = Conv2D(filters,(1,2),kernel_initializer="he_uniform",padding="same")(conv1)
    conv21 = Conv2D(filters,(2,1),kernel_initializer="he_uniform",padding="same")(conv2)
    conv22 = Conv2D(filters,(1,2),kernel_initializer="he_uniform",padding="same")(conv2)

    conv1 = MaxPooling2D()(conv1)
    conv2 = MaxPooling2D()(conv2)
    conv11 = MaxPooling2D()(conv11)
    conv12 = MaxPooling2D()(conv12)
    conv21 = MaxPooling2D()(conv21)
    conv22 = MaxPooling2D()(conv22)

    hidden = concatenate([Flatten()(conv1),
        Flatten()(conv2),
        Flatten()(conv11),
        Flatten()(conv12),
        Flatten()(conv21),
        Flatten()(conv22)])

    x = BatchNormalization()(hidden)
    x = Activation('relu')(x)

    for width in [512,256]:
        x = Dense(width,kernel_initializer="he_uniform")(x)
        x = BatchNormalization()(x)
        x = Activation('relu')(x)

    x = Dropout(0.05)(x)
    outputs = Dense(4,activation="softmax")(x)
    model = Model(inputs,outputs)
    #model.summary()
    model.compile(optimizer="adam",loss="categorical_crossentropy",metrics=["accuracy"])

    return model