train_model.py

import tensorflow as tf
from tensorflow import keras
import numpy as np
from tensorflow.keras.models import Sequential, Model
from tensorflow.keras.layers import Dense, Dropout, Activation, Flatten, Conv2D, MaxPooling2D, Input, Concatenate
from tensorflow.keras.callbacks import TensorBoard
from keras.layers.merge import concatenate
import time

# Loading data
data = np.load("training_data/data.npz",allow_pickle=True)
X = data['arr_0']

Y = data['arr_2']

Z = data['arr_1']

X = X/255.0

Y = Y/255.0

#--------------Optimization--------------
dense_layers = [1]
layer_sizes = [64]
conv_layers = [3]

for dense_layer in dense_layers:
    for layer_size in layer_sizes:
        for conv_layer in conv_layers:
            Name = "{}-conv-{}-nodes-{}-dense-{}".format(conv_layer, layer_size,dense_layer, int(time.time()))
            print(Name)
            # TensorBoard
            tensorboard = TensorBoard(log_dir='logs/{}'.format(Name))

            # Model for image processing
            # First layer
            input1 = Input(shape=X.shape[1:])
            m = Conv2D(layer_size, (3,3))(input1)
            m = Activation("relu")(m)
            m = MaxPooling2D(pool_size=(2,2))(m)

            # Hidden layers
            for l in range(conv_layer-1):
                m = Conv2D(layer_size, (3,3))(m)
                m = Activation("relu")(m)
                m = MaxPooling2D(pool_size=(2,2))(m)
            flatten1 = Flatten()(m)

            # Model for object processing
            # Inputs
            input2 = Input(shape=Y.shape[1:])
            x = Conv2D(32,(3,3),input_shape=Y.shape[1:])(input2)
            x = Activation('relu')(x)
            x = MaxPooling2D(pool_size=(2,2))(x)

            for i in range(2):
                x = Conv2D(32,(3,3))(x)
                x = Activation("relu")(x)
                x = MaxPooling2D(pool_size=(2,2))(x)
            flatten2 = Flatten()(x)

            # Merging models
            #concatenated = concatenate([output1,output2])
            concatenated = Concatenate()([flatten1, flatten2])

            # adding second dense layer
            l = Dense(64+32)(concatenated)
            l = Activation('relu')(l)
            #l = Dropout(0.2)(l)

            l = Dense(64+32)(l)
            l = Activation('relu')(l)
            l = Dropout(0.2)(l)
            output = Dense(4,activation='sigmoid')(l)

            merged_model = Model([input1, input2], output)
            merged_model.compile(loss='binary_crossentropy',
                    optimizer = 'adam',
                    metrics=['accuracy'])

            print(merged_model.summary())
            prev_time = time.time()

            merged_model.fit([X,Y],Z, batch_size=64,epochs=20,validation_split=0.1,callbacks=[tensorboard])
            #model.fit(X,Y, batch_size=64,epochs=20, validation_split=0.1, callbacks=[tensorboard])
            print("Training took: {}".format(int(time.time() - prev_time)))
            merged_model.save("model/{}x{}x{}-CNN.model".format(layer_size, conv_layer, dense_layer+1))