An example of using keras with tf queues, this handle BatchNorm

TFQueueKeras.py

import tensorflow as tf
import numpy as np
import keras
import keras.backend as K
from functools import reduce
from keras.models import Model
import keras.callbacks as cbks
from keras.applications import ResNet50
from keras.layers import Conv2D
import threading
from keras.layers import Layer, Input, InputLayer, Dense
import operator
def prod(factors):
    return reduce(operator.mul, factors, 1)

TRAINING = True
with K.get_session() as sess:
    shp = [10,200,200,3]
    shp1 = [10,7,7,80]
    inp = K.placeholder(shp)
    inp1 = K.placeholder(shp1)
    queue = tf.FIFOQueue(20,[tf.float32,tf.float32],[shp,shp1])
    x1,y1 = queue.dequeue()
    enqueue = queue.enqueue([inp,inp1])
    model = keras.applications.ResNet50(False, "imagenet", x1, shp[1:])
    for i in range(3):
        model.layers.pop()
        model.layers[-1].outbound_nodes = []
        model.outputs = [model.layers[-1].output]
    output = model.outputs[0]  # 7x7
    # Reduce filter size to avoid OOM
    output = Conv2D(32, (1, 1), padding="same", activation='relu')(output)
    output3 = Conv2D(5 * (4 + 11 + 1), (1, 1), padding="same", activation='relu')(output) # YOLO output B (4 + nb_class +1)
    cost = tf.reduce_sum(tf.abs(output3-y1))
    optimizer = tf.train.RMSPropOptimizer(0.001).minimize(cost)

    sess.run(tf.global_variables_initializer())

    def genera(coord,enqueue_op):
        while not coord.should_stop():
            print('ENQUEUE')
            sess.run(enqueue_op,feed_dict={inp:np.arange(prod(shp)).reshape(shp).astype(np.float32),inp1: np.arange(prod(shp1)).reshape(shp1).astype(np.float32)})
        print("DONE for thread")

    coordinator = tf.train.Coordinator()
    threads = [threading.Thread(target=genera,args=(coordinator,enqueue)) for i in range(10)]
    for t in threads:
        t.start()

    for i in range(10): #EPOCH
        for j in range(30): #Batch
            print('RUN')
            optimizer_,s = sess.run([optimizer,queue.size()],feed_dict={K.learning_phase():int(TRAINING)})
            print("Queue size : ", s)

    def clear_queue(queue,threads):
        while any([t.is_alive() for t in threads]):
            _,s = sess.run([queue.dequeue(),queue.size()])
            print(s)
    coordinator.request_stop()
    clear_queue(queue,threads)

    coordinator.join(threads)
    print("DONE")