Using numpy.bincount with array weights

I would like to use bincount to sum arrays, however it supports only doubles. For example this works:

np.bincount([1, 1, 0],weights=np.array([1, 2, 4]))
Out: array([ 4.,  3.])

However I would like to use a dimension 2 array as:

np.bincount([1, 1, 0],weights=np.array([[1,1], [2,2], [4,4]]))
ValueError: object too deep for desired array

The desired output is:

Out: array([[ 4.,  4.],[3., 3.]])

Better explanation after comments:

I want to sum each row of the array into the corresponding index.

With a loop it would be:

Bin=np.zeros(2,2)
for i in [1,1,0]:
    Bin[i]+=a[i]

a is previous 3x2 matrix Есть ли эффективный способ получить этот результат?