StopIteration: generator_output = next(output_generator)

Generators for keras must be infinite:

def subtract_mean_gen(x_source,y_source,avg_image,batch):    while True:        batch_list_x=[]        batch_list_y=[]        for line,y in zip(x_source,y_source):            x=line.astype('float32')            x=x-avg_image            batch_list_x.append(x)            batch_list_y.append(y)            if len(batch_list_x) == batch:                yield (np.array(batch_list_x),np.array(batch_list_y))                batch_list_x=[]                batch_list_y=[] 

The error happens because keras tries to get a new batch, but your generator has already reached its end. (Even though you defined a correct number of steps, keras has a queue that will be trying to get more batches from the generator even if you are at the last step.)

Apparently, you've got a default queue size, which is 10 (the exception appears 10 batches before the end because the queue is trying to get a batch after the end).

As the linked question you provided indicates, Keras Generators have to iterate indefinitely, so you can output elements to your training as long as you want. More info on that on this Github issue.

For that, you must do some modificaiton to your generator like:

def subtract_mean_gen(x_source,y_source,avg_image,batch):batch_list_x=[]batch_list_y=[]while 1: #run forever, so you can generate elements indefinitely    for line,y in zip(x_source,y_source):        x=line.astype('float32')        x=x-avg_image            batch_list_x.append(x)        batch_list_y.append(y)        if len(batch_list_x) == batch:            yield (np.array(batch_list_x),np.array(batch_list_y))            batch_list_x=[]            batch_list_y=[]