from keras import backend as K
from keras.engine.topology import Layer
import numpy as np

class MyLayer(Layer):

 def __init__(self, output_dim, **kwargs):
 self.output_dim = output_dim
 super(MyLayer, self).__init__(**kwargs)

 def build(self, input_shape):
 # Create a trainable weight variable for this layer.
 self.kernel = self.add_weight(name='kernel', 
  shape=(input_shape[1], self.output_dim),
  initializer='uniform',
  trainable=True)
 super(MyLayer, self).build(input_shape) # Be sure to call this somewhere!

 def call(self, x):
 return K.dot(x, self.kernel)

 def compute_output_shape(self, input_shape):
 return (input_shape[0], self.output_dim)

from keras import backend as K
from keras.engine.topology import Layer
import tensorflow as tf

class MyMeanPool(Layer):
 def __init__(self, axis, **kwargs):
 self.supports_masking = True
 self.axis = axis
 super(MyMeanPool, self).__init__(**kwargs)

 def compute_mask(self, input, input_mask=None):
 # need not to pass the mask to next layers
 return None

 def call(self, x, mask=None):
 if mask is not None:
 mask = K.repeat(mask, x.shape[-1])
 mask = tf.transpose(mask, [0,2,1])
 mask = K.cast(mask, K.floatx())
 x = x * mask
 return K.sum(x, axis=self.axis) / K.sum(mask, axis=self.axis)
 else:
 return K.mean(x, axis=self.axis)

 def compute_output_shape(self, input_shape):
 output_shape = []
 for i in range(len(input_shape)):
 if i!=self.axis:
 output_shape.append(input_shape[i])
 return tuple(output_shape)

from keras.layers import Input, Masking
from keras.models import Model
from MyMeanPooling import MyMeanPool

data = [[[10,10],[0, 0 ],[0, 0 ],[0, 0 ]],
 [[10,10],[20,20],[0, 0 ],[0, 0 ]],
 [[10,10],[20,20],[30,30],[0, 0 ]],
 [[10,10],[20,20],[30,30],[40,40]]]

A = Input(shape=[4,2]) # None * 4 * 2
mA = Masking()(A)
out = MyMeanPool(axis=1)(mA)

model = Model(inputs=[A], outputs=[out])

print model.summary()
print model.predict(data)

[[10. 10.]
 [15. 15.]
 [20. 20.]
 [25. 25.]]